Ideal clinical care integrates a health care professional’s clinical experience, individual patient preferences and values, and current best clinical evidence. Every hospitalist has his or her own set of clinical skills and experiences and each patient has his or her own beliefs. The current best clinical evidence, however, is universal. How clinicians apply and explain this evidence to individual patients and integrate the evidence into care plans is the art of practicing evidence-based medicine (EBM). This chapter provides practical guidance on formulating questions and uses clinical examples to discuss how to efficiently and effectively use and search the medical literature.

Caring for patients frequently generates clinical questions. A 1985 study found that general practitioners in an office-based practice formulated two important clinical questions for every three patients examined.¹ Since there are millions of research articles in the world’s literature, finding relevant articles and assessing their quality can be time consuming. If a well-formulated clinical question is posed, the process of finding answers is easier.

There are four parts of a clinical question: the patient population, the intervention being considered, the comparison group, and the measurable outcome. Including these four parts to formulate questions will help focus literature searches.²

A 7-month-old with bronchiolitis was admitted to your service last night. In considering how best to care for her, you formulate the following:

1. 
   

   Patient population (includes patient problem)—in children < 2 years with bronchiolitis

   
   
2. 
   

   Intervention (diagnostic test or treatment), Prognostic factor, or Exposure—do bronchodilators

   
   
3. 
   

   Comparison group—compared with placebo

   
   
4. 
   

   Measurable outcome—reduce hospital length of stay?

Often, developing one critical clinical question will suffice, but a series of questions is sometimes required, especially when inpatients have complex medical problems. Clinical questions should be focused but not overly narrow, since too narrow a question may not have an answer, or may have an answer that does not apply to the individual patient.

In pediatrics, defining the patient population is important. Many hospitalists care for patients ranging in age from infants to young adults, and judgments must be made if studies that do not include your patient’s age will apply. Frequently, a study in adults has been performed and the hospitalist must decide if the study applies to a particular pediatric or adolescent patient.

Sometimes the question may not involve an intervention, and there may therefore be no comparison group to consider. For example, during morning rounds you hear about a 16-year-old female with anorexia nervosa admitted for bradycardia. You ask the question: In a female adolescent with a restrictive eating disorder and heart rate of 35, what is the risk of sudden cardiac death? This question considers the prognosis of a specific population.

The measurable outcome chosen reflects the current clinical concerns, values, and preferences of the clinician and patient. When a child is admitted with right lower quadrant abdominal pain the clinician may be most concerned about how to best evaluate for appendicitis, but the patient and parent may be most concerned about pain control.

A solid comprehension of the disease process or syndrome usually must precede asking specific questions. Background information about pathophysiology, clinical presentation, treatment, complications, and outcomes may be found in print textbooks, review articles, or electronic textbooks. Print textbooks are valuable resources, particularly for well-established information. For example, if a patient presents with a parapneumonic effusion, reading a chapter in a textbook will be helpful in understanding the underlying disease process. A textbook chapter, however, may not contain up-to-date information on management from recently published articles. If you require further background information about parapneumonic effusions, accessing electronic texts may be most helpful.

You begin your reading by reviewing the topic of parapneumonic effusions in children on UpToDate (www.uptodate.com), an easily accessible electronic textbook that is frequently revised and contains good references. If you enter “parapneumonic effusion” in the search line, there is a reference entitled, “Management and prognosis of parapneumonic effusion and empyema in children.” You click on the article and note when the review was most recently updated and when it is due to be revised. In reading the review you remind yourself about the pathogenesis of pleural effusions and the differences between uncomplicated effusions, complicated effusions, and empyemas. There are 82 references, and each reference can be viewed by clicking on the title.

Review articles are another starting place, and Medline searches can be focused to locate review articles by using the “limit” function. For example, if you type parapneumonic effusions.tw (word in title or abstract—see below) and limit the search to review articles and English language, there are 32 hits. The articles include a consensus guideline published in Chest, and Cochrane reviews. Note that a separate search would need to be conducted to find articles that specifically address the definition and management of empyema.

After reminding yourself of definitions and reviewing data, if you have more specific questions, Medline is a good place to begin searching. Medline is a National Library of Medicine database derived from more than 4500 journals and includes over 10 million citations. PubMed, frequently used to access Medline, contains additional basic science and life science articles. Ovid is another means of accessing Medline and is the software referred to in this chapter.

Before you begin searching it is helpful to understand how articles are indexed in Medline. One method of limiting the number of articles you wade through is by attaching suffixes to the keyword. These suffixes allow you to search one journal, search by author, look for words in a title or abstract, or restrict the search by year of publication. Some useful suffixes are listed in Table 3-1. For example, if you search for articles related to bronchiolitis and type bronchiolitis.tw (searches for bronchiolitis as a textword in the title or abstract) you will have 4000 hits. Typing bronchiolitis.ti generates 1900 hits.