It is flu season and patients are coming to the emergency room complaining of fever. Diagnosis of the flu depends on other variables or symptoms though, such as, chills, body aches, and respiratory infection. Just looking at symptoms without testing, what is the probability that the patients actually have the flu? A study by the American Journal of Emergency medicine collected data on 270 emergency room patients having symptoms of the flu, including fever, chills, cough and sore throat. The patients were asked about their symptoms and based on what they described, the physician had to decide if they thought the patient had the flu or not. The goal was to determine how accurate the doctor’s diagnosis were and the probability of them making an accurate diagnosis without an actual test. All 270 patients had at least three of these symptoms and only 42 out of 270 actually had the flu when tested (Dugas, et al., 2015). 42/270 as a simple fraction is 7/45 which is the same as 0.1556 (Bennett, Briggs, & Triola, 2018). So, the doctors in the emergency room have a 16% probability of determining if a patient has the flu without testing (Dugas, et al., 2015).
Rapid influenza diagnostics tests are swabs used to determine if a person actual has the flu when their chief complaint is fever, chills, sore throat, and cough. It takes 15 minutes and is the preferred test for the flu in the ED according to the CDC (CDC.gov, 2018). If 270 patients are seen in the emergency room with some or all of these symptoms, what is the probability that the RIDT test is accurate? The CDC states that if the patient is experiencing a low prevalence of symptoms, being one or two, the probability of the test being a true positive is 39-56% and a false positive is 44-61% (CDC.gov, 2018). So the probability that a person with two or less symptoms has the flu is 105/270- 151/270 for true positive , and 119/270-165/270 probability of a false positive. (Bennett, Briggs, & Triola, 2018) A patient with a high prevalence, having all four symptoms, has a probability of a true positive test result of 90-95% and a false positive probability of 7-14% according to the CDC (CDC.gov, 2018). So, patients with high prevalence of symptoms, having all four, have a probability of 243/270-257/270 for having true positive flu results, and a low probability of 19/270-38/270 for having a false positive result (Bennett, Briggs, & Triola, 2018). The more flu like symptoms a patient has, the higher the probability that it is actually the flu when tested.
Bennett, J., Briggs, W., & Triola, M. (2018). Statistical Reasoning for Everyday Life (5th ed.). Boston: Pearson.
CDC.gov. (2018, March 6). Rapid Diagnostic Testing for Influenza: Information for Clinical Laboratory Directors. Retrieved from Centers for Disease Control: https://www.cdc.gov/flu/professionals/diagnosis/ra...
Dugas, A., Valsamakis, A., Atreya, M., Thind, K., Manchego, P., Faisal, A., & Gaydos, C. (2015). Clinical Diagnosis of Influenza in the ED. The American Journal of Emergency Medicine, 770-775.