More than 40 years ago, Lawrence Weed1 proposed the problem list as the organizing principle of the medical record and emphasized its centrality in medical education:
The instruction of physicians should be based on a system that helps them to define and follow clinical problems one by one and then systematically to relate and resolve them…. [T]he basic criterion of the physician is how well he can identify the patient's problems and organize them for solution.2
A complete, precise, and accurate problem list is essential to guide diagnostic and therapeutic decision making for complex patients and to enable clear communication among clinicians. For clinical educators, the problem list also provides a window on the thought processes of our trainees, as noted by Voytovich3:
Good problem lists and precise terms are important in undergraduate and graduate education because measurement of performance requires explicit structure…. A complete precise problem list provides a skeleton of quintessential data reflecting diagnostic reasoning. In many cases, it is sufficient for judging the quality of that reasoning. “Complete” means that all abnormalities are represented; “precise” means there are no entries that cannot be justified by the existing facts. In other words, everything should be included, and conclusions should be carried as far as possible and no further.
In this article, I will address common deficiencies in the formulation of problem lists and suggest that documentation style and clinical reasoning are closely intertwined. I will argue that the recent trend toward abandoning the problem list in favor of organ checklists and accepting imprecise terminology may contribute to documentation deficiencies and cognitive errors, such as errors of framing, premature closure, and incomplete synthesis. In so doing, I hope to encourage renewed emphasis by clinical teachers on the art of problem formulation and to provoke discussion and further study of the relationships among clinical documentation, reasoning, and evaluation of trainees.
Deficiencies in the Problem List
Common problem list deficiencies include simple omission, incorrect conclusions based on inaccurate or misunderstood data, premature closure, and inadequate synthesis.4 Recent studies have shed light on the frequency of deficient problem lists in medical records. A meta-analysis of 73 studies on the adequacy of hospital discharge summaries, published between 1970 and 2005, found that a median of 17.5% of summaries failed to identify a main diagnosis, while a median of 28% omitted secondary diagnoses.5 A survey of primary physicians' satisfaction with an academic medical center's discharge summaries found that 68% of physicians were dissatisfied with the quality of the summaries, of which they judged discharge diagnoses to be a key ingredient.6
Electronic medical records provide further evidence of the prevalence of deficient problem lists. Although the spread of electronic medical records has facilitated efficient patient care, epidemiologic research, and quality improvement efforts, the quality of an electronic database can be only as robust as the precision and consistency of its diagnostic terms.7 By demanding precise language, the electronic record may promote higher-quality problem lists and help expose deficiencies, provided it does not force unjustified conclusions. A recent study8 exposed the frequency of incomplete problem lists on an academic inpatient service by examining the use of prompts triggered by electronic orders for medications with unique indications for which no appropriate corresponding diagnosis was recorded. For six common medical diagnoses, during a two-month period, prompts were triggered by more than 1,000 problem list omissions, nearly all of which were confirmed by independent review.
The aforementioned studies, two of which were conducted entirely in academic settings, demonstrate a high prevalence of deficient problem lists in the medical record. I suggest that the widespread use of organ-system-based documentation may be one factor contributing to incorrect or incomplete formulation of problem lists; at the least, it reflects a tolerance for imprecise language that is detrimental to rigorous diagnostic thinking.
The Problems With the Organ List
The goal of the diagnostic process is to identify the patient's problems and proceed to explain them at the highest level of resolution that available information allows. Thus, for example, an initially undifferentiated symptom or finding, such as fever, progressively yields its place to bacteremia and subsequently to infectious endocarditis, as further data accrue. In a 2007 editorial, I lamented a trend among medical house officers in recent years to abandon such dynamic lists of problems and progressively refined diagnoses in favor of static lists of organ systems or subspecialties.9 The observation resonated with a number of colleagues around the country, some of whom anecdotally described organ-based notes as “rampant.”
The first data on the prevalence of organ system listing were recently reported from a survey of third-year medical students during their internal medicine clerkship at one U.S. teaching hospital. Upon entering the clerkship, 25% of the students reported that they favored organizing their notes in an organ-based format, a choice they indicated was made mainly on the basis of prior instruction. Their experience during the clerkship influenced the majority toward problem-based notes, yet 52% reported having been instructed by at least one supervisor to adhere to the system-based format.10
However, organ-based progress notes and discharge summaries produce a cumbersome and confusing medical record.9 In the aforementioned survey, medical students using this format were nearly four times more likely than students using problem lists to report they had difficulty deciding where to place some items in the assessment. I put forward here the argument that the disembodied organ list and the thought pattern it represents may predispose trainees to several varieties of faulty or incomplete synthesis, as described below.
Framing error and premature closure
A taxonomy of cognitive diagnostic error has been elaborated over the last two decades.11–13 One of the most important and frequent diagnostic errors is premature closure, the tendency to accept a plausible if unverified diagnosis and end the investigation without sufficient consideration of alternate possibilities. An analysis of autopsy-verified diagnostic errors on an academic medical service found that faulty synthesis, rather than deficient knowledge or diligence, accounted for the majority of misdiagnoses.14 Of the subtypes of defective synthesis identified, premature closure was the most common, followed by faulty context generation, also known as “framing.”
Framing has been defined as “developing a cognitive representation of the patient's problem … [and] establishing a context within which the problem presumably will be solved.”11 Framing error involves “being swayed by subtle wording” to consider a problem from an incorrect perspective.13 I propose that listing issues according to organs or subspecialties, rather than listing problems as the physician perceives them, is likely to introduce framing error and increase the associated risk of premature closure. Further, changing the medical record from a dynamic instrument reflecting progress on a diagnostic trajectory into a static checklist of immutable categories allows the possibility of no closure at all.
The contribution of organ listing to framing error is illustrated by considering a hypothetical patient with dyspnea. Forced to assign this symptom to a single organ system, most clinicians would choose pulmonary, which tends to exclude cardiac disease—as well as other, less common causes—from consideration. Further, when a symptom defies easy organ assignment, trainees are apt to select a subspecialty rubric under which to express the problem; hence, unexplained fever is often labeled infectious disease (ID), which at least temporarily removes from consideration the many noninfectious causes of fever. Equivocal evidence of perhaps trivial infection, such as a mildly abnormal urinalysis, is more easily assigned false blame if the chart insists that the fever has an infectious cause. If a clinician expresses noninfectious fever as ID and goes on to carefully exclude infection as the cause of the fever, then ID is resolved but the fever is not. While the mindful clinician may return to a broader differential diagnosis when no infection is found, framing the initial question as “What infection does this patient have?”—rather than “What is the cause of this patient's fever?”—nonetheless risks premature closure due to the possibility of partial search satisfaction. Search satisfaction “reflects the universal tendency to call off a search once something is found.”11
Diagnostic inefficiency: Vertical versus lateral thinking
I propose that organ listing also promotes diagnostic inefficiency, which is more subtle and doubtless more common than gross diagnostic error. Croskerry12 has developed a particularly extensive lexicon of dispositions toward cognitive diagnostic error. One of these dispositions—“vertical line failure,” or “thinking in silos”—seems to describe organ-based documentation exactly:
[P]redictable, orthodox styles that emphasize economy, efficacy and utility. Though often rewarded, the approach carries the inherent penalty of inflexibility. In contrast, lateral thinking styles create opportunities for diagnosing the rare, unexpected or esoteric. An effective lateral thinking strategy is simply to pose the question: “What else might this be?”12
Trainees sometimes justify the use of the organ list as a checklist to ensure that no aspect of the patient's condition has been overlooked. Organ-system checklists, whether written or mental, may indeed be valuable for generating the differential diagnosis of an unresolved problem. But a checklist in such instances is a means to an end, namely to begin the journey toward diagnosis. As problems and diagnoses are progressively clarified, they, not the organ checklist, should form the assessment and populate the medical record.
Even though checklists are imperative for complex processes, such as supporting a critically ill patient through multiorgan failure,15 they do not replace the need for diagnostic synthesis. Fraser and Greenhalgh16 highlight this distinction between process and problem solving:
“Checklist driven” approaches to clinical care … are important and undoubtedly save lives. But what often goes unnoticed is that such approaches are useful only once the problem has been understood. For the practitioner to make sense of problems in the first place requires intuition and imagination.
For problems with protean manifestations, static lists of organs or subspecialties, each addressed in parallel isolation, cripple the diagnostic imagination. Complex medical illness can be understood only by appreciating the pathophysiologic relationships among multiple affected organ systems. The ability to understand these interactions is the defining skill of the general internist. Consider, for example, a patient with congestive heart failure whose manifestations include pulmonary edema, renal insufficiency, and hepatic congestion—symptoms much more easily understood as a whole than if each organ is considered separately. Weed1 wrote that the author of a note considering each manifestation as a separate issue “is clearly admitting that he does not recognize cardiac failure when he sees it.”
Research supports the view that higher-order problem formulation promotes effective clinical reasoning.17 Bordage18 has connected diagnostic success to elaborated, semantically rich problem statements that place findings in a coherent framework; such statements contrast with the dispersed knowledge structure represented by static, disconnected checklists that fail to identify key associations. Organ listing tends toward the latter extreme, by explicitly avoiding elaboration and obscuring the connections among possibly related findings.
Analysis versus synthesis
To this point, I have focused on organ listing as an impediment to diagnostic progress for unresolved problems. But when a diagnosis is already relatively clear, insistent organ listing may have a regressive effect, obscuring or even deconstructing pathophysiology. An example from my own experience is illustrative: A resident caring for a dyspneic patient—whose problem was persistently framed simply as pulmonary despite overwhelming clinical and radiographic evidence of congestive heart failure—noted that dyspnea and tachycardia occurred with pulmonary embolism and proposed imaging studies for the same, ignoring the obvious, established cause of the patient's symptoms. Synthesis may be abandoned for analysis: Another resident, taking an organ-based approach to a patient with pneumonia, identified the patient's associated fever, hypoxemia, and leukocytosis as distinct infectious, pulmonary, and hematologic issues, respectively. In such instances, firm and satisfying diagnoses are actually undone.
An optical analogy may be helpful in contrasting organ-based and problem-based documentation styles. Organ-based notes, as in these examples, act as a diverging lens; even the clearest image loses its coherence when viewed through it. Notes organized around progressive problem resolution act as a converging lens, collecting parallel streams of information from various sources to cast the problem in sharp focus.
Problem resolution: Static versus dynamic thinking
The value of progressive, accurate diagnostic refinement is at the heart of this discussion. To the thoughtful diagnostician, an unexplained symptom or laboratory finding stands as a challenge, demanding investigation and resolution. Resolving it efficiently, accurately, and as precisely as possible is a source of satisfaction; documenting this progress can only increase the intellectual reward and can light the way to effective therapy and a coherent discharge summary.
Diagnosis and therapy are journeys with direction and a destination. Clinicians must always be aware of their coordinates on this voyage. By “coordinates,” I refer to the level of diagnostic resolution achieved for each issue: beginning with an unexplained symptom or laboratory finding, continuing to a defined even if undiagnosed problem, and arriving at a precise etiologic diagnosis. The well-formulated and refined problem list shows the route clearly—for instance, traveling from fever in an intravenous drug user to Gram-positive coccemia and onward to Staph aureus endocarditis—with an update on testing, therapy, and response given at each step. The organ list, however, is by nature static: Progress notes so organized tend to be more accurately described as status notes. Such a list may easily avoid comment and leave resolution as an open question—for example, ID: begin vancomycin.
On the path to diagnosis, the medical record can serve as a map, with progressively refined problem statements acting as signposts at forks in the road.17 The disembodied organ list, by contrast, is analogous to a packing checklist: It ensures that one has the necessary supplies for the trip, but it does not help one decide where to go, how to get there, or how to determine when one has arrived.
It may be argued that the higher resolution of the problem or diagnosis statement heightens the risk of premature or incorrect closure. Thoughtful essays have questioned the value of discrete and arbitrary diagnostic labels for conditions that exist as continuous variables and have pointed out the harm that can arise from careless application of labels unsupported by evidence in order to satisfy a patient's or payer's cravings for certainty.19,20 When clinicians do not know the answer, we must frankly admit it and reserve the right to say so.21–23 Yet these cautions do not negate the importance of making and expressing a diagnosis when we can. Rather, they underscore the need to express problems or diagnoses as accurately and precisely as possible.23 Rational decisions on further testing and therapy, and appropriate application of literature and guidelines, require open-minded commitment to a working diagnosis. The more precisely a working diagnosis and assumptions are expressed, the more easily they may be subjected to critical examination. By stating explicitly what is known, the physician is forced to acknowledge what is not known, forming a basis for further diagnostic and therapeutic decision making.
Problem formulation in assessing and developing medical knowledge
In addition to promoting diagnostic clarity, thoughtful charting can also help forge the active “connection between the knower and the known” identified by Epstein24 as a key element of mindful practice. Thus, beginning learners may have the most to lose when their clinical coaches tolerate imprecise diagnostic formulations. It is troubling to hear students speak of treating fever with antibiotics or dyspnea with diuretics. Further, the slighting of precise problem formulation may obscure important deficits in medical knowledge, particularly among medical students. One student of mine who favored organ lists, and identified ascending cholangitis simply as GI without further elaboration, turned out to be unable to distinguish among the consequences of obstruction of the cystic, pancreatic, and common bile ducts.
Ironically, even misconceptions can be incorrectly framed. In discussing the inappropriate use of prophylactic acid suppression, typically expressed as GI prophylaxis, students sometimes cite gastroesophageal reflux rather than stress ulcers as the putative indication—they are unaware of precisely what they are trying unnecessarily to prevent. As clinical educators, we therefore owe it to our students to demand documentation in which they clearly express their understanding of each issue and to insist that, to repeat Voytovich,3 “conclusions should be carried as far as possible and no further.”
A Proposed Agenda for Research and Teaching
The phenomenon of organ-based documentation raises several questions that are ripe for further exploration by cognitive psychologists. To begin with, beyond stylistic considerations, does the format of documentation matter? Clinical educators rely heavily on documentation to evaluate our trainees' thinking. Are we correct to assume that organ-based documenters actually think that way, or should we consider that they may convert their thoughts to this form in the belief that it is the preferred method of expression? Do the diagnostic disadvantages shown for other forms of sparse problem formulation apply to the organ-based format as well? In a program with trainees who display diverse documentation habits, systematic comparison of discharge summaries could test the hypothesis that progressive problem listers document diagnostic closure more frequently and more accurately than those who organize their notes around organs or subspecialties. Efficiency could also be compared by examining test-ordering habits and costs of care.
And if organ-based thinking is harmful, as I suggest, how widespread is it really? As I write this article, available data are limited to the aforementioned survey of medical students conducted at a single academic medical center. Data from more institutions would help clarify its prevalence. Some uncertainty also surrounds the organ list's origins and distribution. Checklists have become standard and are of proven value in critical care, where processes are particularly complex. In my own narrow experience—to risk a viral analogy—the use of organ system lists appeared to jump from academic intensive care to general medical services within the last 10 to 15 years. But is the use of such lists in fact limited to current and recent university trainees, or is it also practiced by more experienced physicians in the larger community? And if so, to what effect?
A more challenging line of inquiry concerns factors that help perpetuate organ-based thinking in U.S. teaching hospitals. Here I dare to posit a contributing role for the subspecialty organization of patient care and teaching in academic medical centers. In my institution, the liver, biliary tree, and gastrointestinal lumen are addressed by three distinct consultant groups. Separate services tend to ischemic, dysrhythmic, and failing hearts. Classification by organ or subspecialty may begin in the emergency department with discussion of the appropriate service assignment for the admission, sometimes before initial data collection and problem formulation are even complete. It might be interesting to compare the documentation styles of trainees in quaternary care university centers with those in more primary-care-driven community teaching hospitals to see if structures of knowledge and clinical service organization mirror each other.
These speculations are not meant to devalue subspecialty care, which is indispensable for patients with many diagnoses. Subspecialty teaching helps equip modern learners with extraordinary competence. Trainees seem to know more answers than ever before, but clinical educators must ensure that they retain the capability to think laterally and pose the right questions, lest in their expertise on diseased organs they lose sight of the organism.25
To this end, I encourage clinical educators to renew their insistence that trainees frame and record clinical problems as precisely as possible and continually refine terms as the diagnostic journey progresses. Bowen,26 writing on the teaching of clinical reasoning, reminds us that developing a mental representation of the problem is a critical but often unspoken (and I would add unwritten) early step in the diagnostic process and is a common stumbling block for learners. Alexander27 points out that the traditional clinical skills curriculum of U.S. medical schools is organ-system-based, which equips students rather poorly to approach patients with protean symptoms or disease manifestations; he proposes introducing basic principles of clinical reasoning alongside clinical skills during the second preclinical year. Weed2 may have offered the best prescription when he wrote that the education of a physician
becomes defective not when he is given too much or too little training in basic science … but rather when he is allowed to ignore or slight the elementary definition and the progressive adjustment of the problems that comprise his clinical experience. The teacher who ultimately benefits students the most is the one who is willing to establish parameters of discipline in the not unsophisticated but often unappreciated task of preventing this imprecision and disorganization.
I propose not a new innovation in clinical teaching but instead a renewed emphasis on the fundamental exercise of synthesis. In attending rounds and morning report, formulating a complete list of problems at least, and diagnoses when possible, and developing an understanding of their interrelationships should precede detailed discussion of therapeutic plans. Each problem should be identified as precisely as available data permit and expressed in the most descriptive terms possible. We should ensure that trainees can articulate the level of diagnostic resolution achieved and, hence, clarify what remains to be resolved. If we also insist that the medical record reflect this clarity, the diagnostic and therapeutic process can be illuminated for all involved in the care of the patient.