Steven McGee

Simplifying Likelihood Ratios

Likelihood ratios are one of the best measures of diagnostic accuracy, although they are seldom used, because interpreting them requires a calculator to convert back and forth between “probability” and “odds” of disease. This article describes a simpler method of interpreting likelihood ratios, one that avoids calculators, nomograms, and conversions to “odds” of disease. Several examples illustrate how the clinician can use this method to refine diagnostic decisions at the bedside.

Teaching in the Outpatient Clinic: Practical Tips

ConclusionsOne of the strengths of ambulatory education is the opportunity for learners to interact with patients and for teachers to model what they enjoy most about medicine without the intervening technology common to hospital wards. When medical students from the graduating class of 1990 were asked in a national survey what would make the specially of internal medicine more attractive as a career, the most common suggestion was to increase the ambulatory experience and the connections with patients that such an experience provides.More research is necessary to identify which teaching techniques are effective in the clinic.3 Meanwhile, this article makes practical suggestions based on what is known about effective teachers and their behaviors. When teachers ask questions, present general rules, and model interactions, they create brief opportunities for teaching in an otherwise hectic day. Not only do learners recall these general rules, they subsequently want to emulate the teachers caring attitude toward patients and organized approach to problem solving. Asking questions and modeling interactions help teachers share themselves and their love of medicine with their learners.

Does This Patient Have a Hemorrhagic Stroke?: Clinical Findings Distinguishing Hemorrhagic Stroke From Ischemic Stroke

CONTEXT The 2 fundamental subtypes of stroke are hemorrhagic stroke and ischemic stroke. Although neuroimaging is required to distinguish these subtypes, the diagnostic accuracy of bedside findings has not been systematically reviewed. OBJECTIVE To determine the accuracy of clinical examination in distinguishing hemorrhagic stroke from ischemic stroke. DATA SOURCES MEDLINE and EMBASE searches of English-language articles published from January 1966 to April 2010. STUDY SELECTION Prospective studies of adult patients with stroke that compared initial clinical findings with accepted diagnostic standards of hemorrhagic stroke (computed tomography or autopsy). DATA EXTRACTION Both authors independently appraised study quality and extracted relevant data. DATA SYNTHESIS Nineteen prospective studies meeting inclusion criteria were identified (N = 6438 patients; n = 1528 [24%] with hemorrhage stroke). Several findings significantly increase the probability of hemorrhagic stroke: coma (likelihood ratio [LR], 6.2; 95% confidence interval [CI], 3.2-12), neck stiffness (LR, 5.0; 95% CI, 1.9-12.8), seizures accompanying the neurologic deficit (LR, 4.7; 95% CI, 1.6-14), diastolic blood pressure greater than 110 mm Hg (LR, 4.3; 95% CI, 1.4-14), vomiting (LR, 3.0; 95% CI, 1.7-5.5), and headache (LR, 2.9; 95% CI, 1.7-4.8). Other findings decrease the probability of hemorrhage: cervical bruit (LR, 0.12; 95% CI, 0.03-0.47) and prior transient ischemic attack (LR, 0.34; 95% CI, 0.18-0.65). A Siriraj score greater than 1 increases the probability of hemorrhage (LR, 5.7; 95% CI, 4.4-7.4) while a score lower than -1 decreases the probability (LR, 0.29; 95% CI, 0.23-0.37). Nonetheless, many patients with stroke lack any diagnostic finding, and 20% have Siriraj scores between 1 and -1, which are diagnostically unhelpful (LR, 0.94; 95% CI, 0.77-1.1). CONCLUSION In patients with acute stroke, certain findings accurately increase or decrease the probability of intracranial hemorrhage, but no finding or combination of findings is definitively diagnostic in all patients, and diagnostic certainty requires neuroimaging.

Use of Corticosteroids in Treating Infectious Diseases

Clinicians have generally avoided prescribing corticosteroids for active infection because of their known immunosuppressive effects and concern about long-term complications. We conducted a review of the published randomized, double-blind trials comparing corticosteroids and placebo in infections. Except in some trials of viral infections, sore throat, and cerebral cysticercosis, all patients also received active antimicrobial agents in addition to placebo or corticosteroids. For patients with bacterial meningitis, tuberculous meningitis, tuberculous pericarditis, severe typhoid fever, tetanus, or pneumocystis pneumonia with moderate to severe hypoxemia, treatment with corticosteroids improved patient survival (group 1 infections). For patients with bacterial arthritis, corticosteroids were also beneficial and reduced long-term disability (group 2 infections). For about a dozen other infections, corticosteroids significantly relieved symptoms (group 3 infections), and clinicians should consider using them if symptoms are substantial. Corticosteroids were harmful in 2 infections, viral hepatitis and cerebral malaria (group 5 infections). We conclude that corticosteroids are beneficial and safe for a wide variety of infections, although courses longer than 3 weeks should be withheld from patients with concomitant human immunodeficiency virus infection and low CD4 counts.

Bedside Teaching Rounds Reconsidered

In his 1905 farewell address to the staff of Johns Hopkins Hospital, Sir William Osler reflected on 16 years of accomplishments in the field of medical education, concluding that his greatest achievement was “[teaching] medical students in the wards, as I regard this as by far the most useful and important work I have been called upon to do.”1 Photographs of Osler examining patients and teaching others (Figure) reveal his systematic approach and its command on his trainees’ attention.2 Osler has become a model for modern teachers, although few of them currently regard bedside teaching as their greatest achievement. Most surveys today show that teachers spend less than 25% of their teaching time at the bedside, instead preferring conference rooms or the hallway outside patients’ rooms as the place of instruction.3,4 Barriers to bedside teaching—mostly absent during Osler’s time—are insufficient time to teach, dependence of diagnosis on technology, obstacles created by infection control, and distractions from clinical responsibilities at distant computer stations. Also, many teachers find bedside teaching inherently difficult: normal clinical activities frequently interrupt all but the shortest of prepared bedside lessons, and spontaneous questions from patients and learners at the bedside are often unanswerable, at least immediately, and may fall outside the teacher’s expertise. Despite these challenges, there are compelling reasons to continue bedside teaching. Instead of increasing patient anxiety, bedside rounds can help calm patients, making them more likely to view physicians and trainees favorably and to be more confident that their medical problems are fully understood.5,6 Even in critically ill patients, measurements of vital signs and norepinephrine levels during bedside presentations show no signs of physiologic stress.7 Once students and house officers experience bedside teaching, they tend to prefer such rounds for future instruction, commenting that bedside rounds provide them their only opportunity to see teachers interact with patients, learn physical diagnosis, and reinforce the perspective that patients are not abstract diseases or hosts but instead unique persons.4,6 Many patients experience satisfaction and pleasure in helping educate others, which can establish a special bond between them and their medical team. Given these advantages, how can we address the challenges of bedside teaching and make it routine for all trainees? First, the oral case presentation must be brief. With the widespread availability of the electronic medical record, many teachers now review much of the patient’s history and testing before bedside rounds, making the traditional long presentation not only unnecessary but wasteful of the limited time available. More concise presentations, emphasizing just the essential findings and current plans, help everyone focus on the patient’s principal problems and leave time for teachers to clarify details and instruct trainees. Of the many ways to abbreviate presentations,8 one format appears in eAppendix 1 in the Supplement. Second, clinical teachers should recognize that there are two distinct kinds of instruction: one best taught in the classroom, and another that can be taught only at the bedside. The first type includes complex discussions of pathophysiology and management, subjects difficult to address fully at the bedside but more easily conveyed in classrooms, where trainees can sit down and teachers can take advantage of handouts, white boards, and fewer interruptions. The natural subjects of bedside teaching, in contrast, are communication, professionalism, and clinical skills. Trainees closely observe all of the teachers’ behaviors—how they handle introductions, address patient concerns, elicit key details, ask permission to examine, explain symptoms, avoid jargon, respect the medical team, and bond with the patient—subsequently borrowing for their own practice the empathetic and therapeutic behaviors and discarding the unhelpful ones. Instruction on clinical skills is more explicit and often focuses on the fundamentals of physical diagnosis, but teachers should be aware that there are two types of physical findings—those with technological surrogates and those without such surrogates. (A physical finding has a technological surrogate if the diagnostic standard for a disorder is a laboratory test or clinical image; for example, the physical findings of pneumonia have a technological surrogate—chest radiography— whereas the skin findings of cellulitis do not.) To address skepticism among some trainees who are enamored by technology and cynical about physical diagnosis, a fundamental lesson during bedside rounds is that the diagnosis of many clinical problems, despite modern testing, still depends primarily on what the clinician sees, hears, and feels (BOX). Third, because few teachers can speak eloquently about every clinical skill without preparation and because the time available to teach is usually meager, one effective strategy is to compose “teaching scripts,” minilectures that the teacher memorizes and then uses during bedside rounds. Examples of scripts are “Distinguishing cellulitis from dermatitis,” “Predicting prognosis in pneumonia,” or “Differentiating hyperthermic syndromes” (eAppendix 2 in the Supplement presents one script, “Diagnosing Parkinson disease”). Sometimes the problems of a specific patient prompt a particular script, but other times the teacher creates the teaching opportunity by posing a question, awaiting an answer, and then delivering the lesson. Fourth, we should celebrate the triumphs of bedside examination. In my own institution during recent A PIECE OF MY MIND

Percussion and physical diagnosis: Separating myth from science

There are three percussion sounds, which are easily distinguishable by objective measures: tympany (heard with percussion over the intestines), resonance (heard over the normal lung), and dullness (heard over the liver or thigh). The percussion sound that is produced reflects the ease with which the body wall vibrates, which in turn is influenced by many variables, including the strength of the stroke, the condition and state of the body wall, and the underlying organs. Underlying organs or disease may cause dullness to occur at distant sites. There is good interobserver agreement among clinicians with regard to calling a particular percussion sound dull, resonant, or hyperresonant. In contrast, there is very poor interobserver agreement among clinicians using percussion to measure the span of a particular organ. The use of comparative percussion can detect most large pleural effusions, but this method is able to detect only a few pneumonias. Shifting dullness is a reliable and fairly accurate sign for the detection of ascites. Both of these techniques can still be recommended after a review of the literature. Topographic percussion (e.g., using percussion to locate the heart, liver, and spleen borders or dimensions) has poor reproducibility, is significantly inaccurate in many patients, and has little clinical utility; it should be abandoned. Its fundamental principle--that sound waves penetrate only several centimeters of tissue, resulting in a note reflecting abnormalities only in this layer of tissue--is incorrect. Auscultatory percussion offers no advantage over conventional percussion, with the possible exception of auscultatory percussion of the shoulder. Auscultatory percussion should be abandoned as a bedside diagnostic technique.

Diagnosing and Treating Dizziness

Dizziness is a common presenting concern in primary care practice. The most useful diagnostic approach in distinguishing different types of dizziness is a thorough history and physical examination; additional tests are rarely necessary. Effective treatments exist for many causes of dizziness, and these treatments are often accomplished in the clinic or at home without the need for medication.

Therapy for Cellulitis

In Reply Dr Goldman and colleagues highlight aspects of the phylogenetic analyses used to investigate HIV transmission events in the PARTNER study.1 Previous studies have taken advantage of the now-discontinued Roche 454 deepsequencing platform to obtain sequence reads of sufficient length to allow reliable phylogenetic analyses of minority viral species. We have been conducting work to optimize the Illumina deep-sequencing platform to perform an analysis of minority species in couples in our study. The reconstruction of HIV haplotypes presents notorious technical and interpretative challenges when applied to the short sequence reads currently obtained by Illumina. We are also using conventional limiting dilution techniques to obtain single and near full-length genomes using established methods.2 These further analyses will be submitted for publication once completed. Also, Goldman and colleagues propose that some env pairwise genetic distances in samples from the PARTNER study were similar to those of samples found to be linked in another study.3 The proposed comparison of genetic distances is complicated by the fact that the sequences in the PARTNER study were considerably longer (2000 base pairs) than those reported in the other study (approximately 516 base pairs). Nonetheless, as shown in eTable 2 in the article Supplement, the median pairwise distance of env control sequences was at least 5 times lower than the median pairwise distance of the partners’ env sequences. When considering sequences falling within the upper limit of the previously reported range,3 the env phylogeny did not support linkage. Detailed analyses of the env sequences were made available to selected expert reviewers from JAMA and deemed robust. All phylogenies will be submitted for publication once study is completed. Goldman and colleagues are correct that phylogenetic analyses of putative transmission events should include control sequences drawn from epidemiologically relevant settings and take into account time since seroconversion.4 These factors were taken into account in the PARTNER study. The study design was such that patients were sampled never later than 6 to 8 months from seroconversion. Constraints dictated by the terms of the ethical approvals and need to protect patients’ confidentiality mean that the phylogenetic investigations must not reveal the geographic origin of the specimens undergoing analysis. Although we recognize the importance of disclosing to public scrutiny our detailed analyses, the confidential data we hold in this respect are entirely consistent with the reported conclusions of the PARTNER study. We are confident that clinicians are able to interpret the data and counsel patients appropriately, taking into account individual circumstances and tolerance of any risk, however small.