Evidence-based Approach: Introduction

Acquiring solid history-taking and physical examination skills, the ability to use and interpret them in the right way are essential for physicians.

The literature shows the physical examination, in general, is considered in decline while the use of laboratory and imaging testing has markedly increased (1-3). Indeed, physicians tend to over-rely on the test results, instead of history and physical examination findings. A reason for this trend may be physicians’ lack of knowledge or confidence (4,5). However, not every institution has the optimal resources and even if they have, extensively testing every patient for every disease possible is not cost-effective or free of complications (5). Therefore, acquiring solid history-taking and physical examination skills, the ability to use and interpret them in the right way are still essential for physicians.

Learning and performing history-taking and physical examination techniques is one thing, applying the findings to reach a diagnosis is another (6). Early learners tend to focus on mastering the skills itself so much that often they may fail to notice how to utilize it, its strengths and weaknesses. iEM Education Project’s new series “Evidence-based Approach to History-taking and Physical Examination” aims to support this learning gap.

In a way, evidence-based history-taking and physical examination challenge traditional habits in an attempt to curate them; gives us the information needed to abandon the invalid techniques and nourish the beneficial ones. However, interpreting the findings relies on understanding the evidence. Therefore, before analyzing each disease from the perspective of evidence-based diagnostic skills, we need to review the biostatistical terms such as pre-test probability, sensitivity and specificity, positive and negative likelihood ratios (LR).

Below is a simple reminder about how to interpret these values. You may refer to the links provided to reach more information.

Pre-test Probability

Pre-test probability is the probability of the disease before implementing any results (7,8). In other words, it is how likely the physician thinks a patient with a chief complaint may have a specific condition. There are three main ways to estimate pretest probability; first, prevalence studies; second, validated clinical prediction rules; and third, physicians’ gestalt based on their own clinical experience (9).

Sensitivity

Sensitivity is a feature (symptom, sign or test) with a high sensitivity is positive more frequently in patients compared to healthy population and selects patients accurately when it is positive (positivity in disease) (7). Therefore, when a highly sensitive feature is absent, the probability of the disease decreases. (SnNout = a Sensitive test, when Negative, rules out disease) (7).

Specificity

Specificity is a feature (symptom, sign or test) with a high specificity is negative more frequently in the health population compared to patients and selects healthy people accurately when it is negative (negativity in health) (7). Therefore, when a highly specific test is positive, the probability of the disease increases. (SpPin = a Specific test, when Positive, rules in disease) (7).

Positive Likelihood Ratio (LR+)

LR+ describes how the probability of a disease changes when a feature (symptom, sign or test) is present (10).

      • If LR+ > 1, the presence of the feature, increases the probability of the disease. The bigger the LR+, the more strongly it favors the diagnosis.
      • If LR+ = 1, the presence of the feature does not change the probability. Therefore, it does not have diagnostic value. 
      • If LR+ = 0-1, the presence of the feature decreases the probability of the disease. The smaller the LR+, the more strongly it opposes the diagnosis (10).

Negative Likelihood Ratio (LR-)

LR- describes how the probability of a disease changes when a feature (symptom, sign or test) is absent (10).

      • If LR- > 1, the absence of the feature, increases the probability of the disease. The bigger the LR-, the more strongly it favors the diagnosis.
      • If LR- = 1, the absence of the feature does not change the probability. Therefore, it does not have diagnostic value.
      • If L- = 0-1, the absence of the feature decreases the probability of the disease. The smaller the LR-, the more strongly it opposes the diagnosis (10).

How to combine all?

In the traditional sense, the pretest probability is used to mean the prevalence of a disease before ordering a test (8). Basically, it is another way of saying physicians used to combine symptoms and signs intuitively, based on their experience to reach a pretest probability. However, evidence-based medicine encourages the physicians and the literature to reflect on the practice, break it into pieces and review the individual and collective value of each part. Accordingly, each individual feature from history or examination can be considered “tests.” (8)

You may think reviewing the value of each feature from the history and physical examination is mentally exhausting. Validated clinical prediction rules are here to help! Similar to the traditional sense, but in an evidence-based and standardized way, the validated clinical prediction rules combine some elements to reach a more straightforward calculation of pretest (9).

Overall, interpreting the findings is as important as performing the skill itself. Interpretation requires biostatistical knowledge as much as clinical ability. When applied analytically, history-taking and physical examination can safely accelerate the diagnostic process and limit overtesting (5).

References and Further Reading

  1. Smith-Bindman, R., Miglioretti, D. L., & Larson, E. B. (2008). Rising use of diagnostic medical imaging in a large integrated health system. Health Affairs, 27(6), 1491-1502.
  2. O’Sullivan, J. W., Stevens, S., Hobbs, F. R., Salisbury, C., Little, P., Goldacre, B., … & Heneghan, C. (2018). Temporal trends in use of tests in UK primary care, 2000-15: retrospective analysis of 250 million tests. British Medical Journal, 363, k4666.
  3.  Bergl, P., Farnan, J. M., & Chan, E. (2015). Moving toward cost-effectiveness in physical examination. The American Journal of Medicine, 128(2), 109-110.
  4. Cook, C. (2010). The lost art of the clinical examination: an overemphasis on clinical special tests. The Journal of Manual & Manipulative Therapy, 18(1), 3.
  5. Greenberg, J., & Green, J. B. (2014). Over-testing: why more is not better. The American Journal of Medicine, 127(5), 362-363.
  6. Chi, J., Artandi, M., Kugler, J., Ozdalga, E., Hosamani, P., Koehler, E., … & Verghese, A. (2016). The five-minute moment. The American Journal of Medicine, 129(8), 792-795.
  7. McGee, S. (2018). Evidence-based Physical Diagnosis (4th Ed., Kindle Ed.). Philadelphia: Elsevier.
  8. Parikh, R., Parikh, S., Arun, E., & Thomas, R. (2009). Likelihood ratios: clinical application in day-to-day practice. Indian Journal of Ophthalmology, 57(3), 217.
  9. Shaneyfelt, T. (2012). Diagnostic Process. [Online Lecture]. Retrieved April 25, 2019 from https://www.youtube.com/watch?v=6qgnrXELoo4.
  10. McGee, S. (2002). Simplifying likelihood ratios. Journal of General Internal Medicine, 17(8), 647-650.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.