Date: July 13th, 2020

Guest Skeptic: Dr. Justin Morgenstern is an emergency physician and the creator of the excellent #FOAMed project called He has a great new blog post about increasing diversity in medicine using something called the BSAP approach and an interesting Broome Doc podcast with Dr. Casey Parker called EBM 2.0.  

Reference: Ebell et al. Accuracy of Signs and Symptoms for the Diagnosis of Community‐acquired Pneumonia: A Meta‐analysis. AEM July 2020

Case: A 67-year-old woman with no previous health problems presents with fever, cough, and myalgias. You are working with a medical student on their very first rotation, and you want to spend some time teaching them about the history and physical exam. However, being an evidence-based medicine enthusiast, you wonder what aspects of the patient’s presentation are going to be truly helpful in making a diagnosis. 

Background: Depending on the time of year, fever and cough can be one of the most common presentations seen in the emergency department. It is important not to miss pneumonia in the sea of viral illnesses. We have covered various aspects of this issue a number of times on the SGEM:

  • SGEM#287: Difficult to Breathe – It Could Be Pneumonia
  • SGEM#286: Behind the Mask – Does it need to be an N95 mask?
  • SGEM#263: Please Stop, Prescribing – Antibiotics for Viral Acute Respiratory Infections
  • SGEM#216: Pump It Up – Corticosteroids for Patients with Pneumonia Admitted to Hospital
  • SGEM#120: One Thing or Two for Community Acquired Pneumonia?

Antibiotic overuse is a significant problem, and ordering chest x-rays (CXR) on everyone is inefficient, expensive, and adds potentially unnecessary risk from radiation. Thus, it is important to know how accurate the history and physical exam is for identifying patients with pneumonia.

A prior meta-analysis demonstrated that the combination of normal vital signs and normal lung exam effectively rules out pneumonia (Marchellow eat al JABFM 2019), and that a physician’s overall clinical impression is moderately accurate (Dale et al BrJGP 2019).

However, there has not been a meta-analysis looking at the evidence for individual signs and symptoms for pneumonia in the last decade.

Clinical Question: What is the accuracy of individual signs and symptoms for diagnosing community acquired pneumonia?

Reference: Ebell et al. Accuracy of Signs and Symptoms for the Diagnosis of Community‐acquired Pneumonia: A Meta‐analysis. AEM July 2020

  • Population: Adolescents and adults presenting with symptoms of respiratory infection or clinically suspected pneumonia in the outpatient setting
  • Intervention: Any clinical sign or symptom (including vital signs) for pneumonia
  • Comparison:
  • Outcome: Radiologically confirmed pneumonia (using CXR as the gold standard)

Dr. Mark Ebell

This is an SGEMHOP episode which means we have the lead author on the show.  Dr. Mark Ebell is a Family Physician and Professor at the University of Georgia in Athens. He is a co-founder of POEMs, editor-in-chief of Essential Evidence, deputy editor of American Family Physician, and co-host of the podcast Primary Care Update and POEM of the Week 

Authors’ Conclusions: While most individual signs and symptoms were unhelpful, selected individual signs and symptoms are of value for diagnosing CAP. Teaching and performing these high value elements of the physical examination should be prioritized, with the goal of better targeting chest radiographs and ultimately antibiotics.

Quality Checklist for Systematic Review Diagnostic Studies:

  1. The diagnostic question is clinically relevant with an established criterion standard. Unsure.
  2. The search for studies was detailed and exhaustive. Yes
  3. The methodological quality of primary studies were assessed for common forms of diagnostic research bias. Yes
  4. The assessment of studies were reproducible. Yes
  5. There was low heterogeneity for estimates of sensitivity or specificity. Yes/No
  6. The summary diagnostic accuracy is sufficiently precise to improve upon existing clinical decision-making models. No

Key Results: They identified 16 studies that met their inclusion and exclusion criteria. Seven studies were based in the emergency department and nine in a primary care setting. The number of participants ranged from 52 to 2850. The mean age ranged from 32 to 62 years, and between 48% and 60% of the participants were female.

A CXR was used as the gold standard in all studies. The risk of bias was assessed as low in 12 studies and moderate in five. The prevalence of pneumonia was 10% in the primary care studies and 20% in emergency department studies.

No individual sign or symptom was good enough to independently rule in or rule out pneumonia.

The most helpful indicator was “overall clinical impression”, with a positive likelihood ratio of 6.32 (the highest of any finding) (95% CI 3.58-10.5) and a negative likelihood ratio of 0.54 (95% CI 0.46-0.64).

Although a number of symptoms and signs were associated with pneumonia, the low positive likelihood ratios – generally less than 2 – mean that none of these factors are even close to diagnostic on their own. Examples include subjective fever, dyspnea, chest pain, dullness to percussion, crackles, confusion, and toxic or ill appearance. The negative likelihood ratios were even less helpful. We will include a full table with the results in the show notes.

The finding with the best test characteristic to rule in pneumonia was egophony, with a positive likelihood ratio of 6.17 (95% 1.34-18.0) when present, although the negative likelihood ratio was only 0.96 (0.93-0.99)

The absence of any abnormal vital sign was the best finding for ruling out pneumonia, with a negative likelihood ratio of 0.25 (95% CI 0.11-0.48)

1. Exclusions: You excluded patients from skilled nursing facilities, with chronic lung disease, and immunosuppressed patients. From a pure diagnostic standpoint, that makes sense. However, these are probably the patients in whom it’s most important not to miss a diagnosis of pneumonia. Based on your results, how do you approach the diagnosis in these patients?

2. Other Databases: You limited your search to the Medline databases, whereas we often see systematic reviews search multiple databases to ensure the results aren’t biased by missing published studies. Can you explain for the listeners why a researcher might decide to search one database versus multiple, and whether you think it could significantly affect the results?

3. Imperfect Gold Standard: The signs and symptoms were compared to CXR. We know that a CXR is less accurate in diagnosing CAP than a CT scan. How do you think that could have impacted the results?

4. Prevalence and Possible Selection Bias: Perhaps it is just the community I work in, where everyone wants their viral illness checked in the emergency department, but a 20% prevalence of pneumonia in the emergency department seems quite high to me. Could this represent selection bias, and if so how might that impact the results?

5. Spectrum Bias: In general, these studies included patients in whom the clinician suspected pneumonia, and so presumably are a sicker cohort than all comers with cough. The negative likelihood ratios would probably look better if we included all comers, and we might be misled into over-testing if we try to apply these results to every patient presenting with a cough.

6. Verification Bias: You mention in your methods that you only included studies in which imaging was either performed on all patients, or all high-risk patients with a random sampling of low risk patients, in order to avoid verification bias. Can you explain verification bias to our listeners, and why it might be important when considering this type of literature?

7. Sensitivity vs. Specificity: The only finding with a moderate sensitivity for ruling out pneumonia was the absence of any abnormal vital signs. I worry that people will hear that result and interpret it as if the patient has an abnormal vital sign, they must get imaging. However, the specificity is going to be pretty low – basically every influenza patient is mildly tachycardic. Can you talk about sensitivity, specificity, and how these numbers actually drive your clinical practice?

8. Limited Utility vs. No Utility: It would be pretty easy to look at these numbers and get a little nihilistic. Is the physical exam even necessary? However, there is a difference between a single criterion having limited impact independently, and it having no impact at all. Presumably, the overall clinician’s impression – which was the most accurate finding – included many of these individual findings, so they may add up to more than the sum of their parts.

9. Clinically Significance : A positive CXR does not mean a patient has a bacterial pneumonia. Prescribing antibiotics to a patient with a viral pneumonia is unlikely to have a patient-oriented outcome (POO). Do you think this disease-oriented outcome (DOO) and not a POO is a problem?

10. Are All Clinicians the Same? Overall clinical judgement was the most accurate for diagnosing pneumonia, but I wonder whether all clinicians are equally good. First, do we know what level of training the participants in these studies were. Second, do you think there are ways that we can improve our own clinical judgement when it comes to pneumonia?

Comment on Authors’ Conclusion Compared to SGEM Conclusion: We agree with the authors’ conclusion that most individual signs and symptoms are unhelpful on their own, but there are a few high value findings, like normal vital signs, or egophony. These findings can be used to teach the physical exam and may help make better decisions about imaging and antibiotic use.

SGEM Bottom Line: No individual sign or symptom is good enough to either rule in or rule out community acquired pneumonia. Decisions should be made based on the clinician’s overall judgement. However, individual clinicians may want to improve their expertise by reviewing radiology results, and following patients up clinically, so that they can learn from their mistakes over time.

Case Resolution: You review the entire history and physical exam with your student, as everyone needs to learn the basics. You explain to the patient that based on your clinical expertise, you think it is unlikely that they have a pneumonia, and so they don’t need to be exposed to CXR at this time. However, you explain to the patient that no test is perfect, so if she is getting worse and is worried that you may have missed a pneumonia, she should come back for a recheck.

Dr. Justin Morgenstern

Clinical Application: Depending on where you are in your career, reviewing these numbers may help you develop the expertise required to accurately diagnose pneumonia, although clinical diagnosis alone will never be perfect.

What Do I Tell the Patient? Based on the symptoms you have, your normal vital signs, and the fact that your lung exam is normal, I think it is very unlikely that you have a pneumonia today, so we don’t have to expose you to an CXR. However, we can never be 100% certain, so if you are getting worse, please come back so we can recheck you.

Keener Kontest: Last weeks’ winner was Matt Corey a PA from Phoenix, Arizona. He they knew Elliot Grossbard (Genentech scientist) is reported to have said “We do not know how another trial would turn out, and if we do not come out ahead, we would have a terribly self-inflicted wound….[Another study] may be a good thing for America, but it wouldn’t be a good thing for us.”   Linda Marsa 1997 and Jeanne Lenzer BMJ 2002

Listen to the SGEM podcast to hear this weeks’ question. Send your answer to with “keener” in the subject line. The first correct answer will receive a cool skeptical prize.

SGEMHOP: Now it is your turn SGEMers. What do you think of this episode on the diagnostic accuracy of signs and symptoms for community acquired pneumonia? Tweet your comments using #SGEMHOP. What questions do you have for Mark and his team? Ask them on the SGEM blog. The best social media feedback will be published in AEM.

Also, don’t forget those of you who are subscribers to Academic Emergency Medicine can head over to the AEM home page to get CME credit for this podcast and article. We will put the process on the SGEM blog:

  • Go to the Wiley Health Learning website
  • Register and create a log in
  • Search for Academic Emergency Medicine – “July”
  • Complete the five questions and submit your answers
  • Please email Corey ( with any questions

Remember to be skeptical of anything you learn, even if you heard it on the Skeptics’ Guide to Emergency Medicine.