Pediatric Pneumonia Podcast

Diagnosing Pediatric Pneumonia: Challenges, Overlapping Clinical Symptoms, and Clinical Prediction Models

Todd A. Florin, MD, MSCE; Mark I. Neuman, MD


May 04, 2023

This transcript has been edited for clarity. For more episodes, download the Medscape app or subscribe to the podcast on Apple Podcasts, Spotify, or your preferred podcast provider.

Todd Florin, MD: Hello. I am Dr Todd Florin, and I am an associate professor at Northwestern University's Feinberg School of Medicine and the director of research for the Division of Emergency Medicine at Ann and Robert H. Lurie Children's Hospital of Chicago. Welcome to Medscape's InDiscussion series on Pediatric Pneumonia. Today, we'll discuss the diagnosis of pediatric pneumonia, a topic that seems simple but becomes challenging as we peel back the layers of the onion. This is largely because the clinical presentation of pneumonia in children overlaps with so many other common respiratory conditions, such as asthma, bronchiolitis, and viral upper respiratory infections. We'll discuss both clinical and radiographic approaches to the diagnosis of pneumonia in children. First, let me introduce my guest, Dr Mark Neuman. Dr Neuman is an associate professor of pediatrics and emergency medicine at Harvard Medical School and the director of research in the Division of Emergency Medicine at Boston Children's Hospital. He has been conducting research in pediatric pneumonia for the past 2 decades. Welcome to InDiscussion.

Mark Neuman, MD: Hi, Todd. How are you?

Florin: Great to see you. Mark, you and I have been working in this field for quite a while now. What do you think are some of the most exciting changes in care that you've seen over your time?

Neuman: One of the best things I've seen is the development of clinical decision rules to help optimize patient care. One of the areas in which there's been huge advancements is in the care of the febrile infant. I remember when I was a fellow training here in Boston, we used to do a full sepsis workup on all infants under 90 days of age, and those kids were admitted to the hospital often for up to 3 days. Now, as a result of clinical decision rules, we typically only perform a lumbar puncture (LP) — or a spinal tap — on kids older than 30 days. We're even moving that range down quite a bit.

Florin: I definitely have seen that changing in my career as well. I feel that it's a good link to the discussion that we're going to have today about how the approach to diagnosis and management to common pediatric infections has really changed as we've learned more about these infections. That leads us directly into our conversation today about the diagnosis of pediatric pneumonia. Pneumonia is really a broad term. We see it used in lots of different ways, some more specific (pus and consolidation in the lung) and others more general (any findings of lower respiratory tract infection in a kid with a fever). What do you think are the best clinical features to predict the presence of a radiographic pneumonia for something that you can see on a chest radiograph?

Neuman: I think the important thing with respect to clinical signs and symptoms of pneumonia is that there's no single sign or symptom that's highly accurate for the diagnosis of pneumonia. The typical scenario of a child who comes into the emergency department with a high fever, a productive cough on exam, maybe hypoxic, working to breathe, and having focal rales is a rare phenomenon. Most children present with a variety of signs and symptoms, but I think you can't really hang your hat on a single sign or symptom to predict pneumonia. Some studies that we and other groups have done found certain findings are more likely to be associated with pneumonia in children, such as the presence of fever, fever duration, decreased breath sounds, having rales or crackles (particularly if they're focal in nature), and hypoxia. I think the biggest takeaway is that no single sign or symptom is highly accurate. With that said, in a meta-analysis that we've done, we found that hypoxia and increased work of breathing were the two findings that were more likely to be associated with pneumonia and that altered your risk of diagnosing pneumonia more than any other sign or symptom. Wheezing was associated with not having radiographic pneumonia.

Florin: When that meta-analysis came out, which was published in the Journal of the American Medical Association (JAMA), I found that really fascinating because I think that traditionally tachypnea has been considered a diagnostic criteria for pneumonia, not only in the US but worldwide. Can you speak a little bit about what you found with regard to the role of tachypnea in diagnosing pneumonia?

Neuman: Tachypnea is interesting. Tachypnea, or rapid breathing, is often associated with pneumonia. However, many of the children who are being considered for having pneumonia also are tachypneic, which makes tachypnea a tough finding to say whether it's associated with pneumonia. Around the world, in resource-poor settings, tachypnea is the criteria that's used to define pneumonia in those settings. However, because tachypnea is associated with many other respiratory illnesses like asthma, bronchiolitis, even with upper respiratory infections, it is not commonly associated with pneumonia with respect to differentiating pneumonia from other types of infections or respiratory illnesses.

Florin: I think that's important for our listeners to know. I also want to flesh out the hypoxemia criteria. At what oxygen saturation do we see that risk of radiographic pneumonia go up?

Neuman: It's a challenging question to answer. In that meta-analysis, there were many studies that were done that used different thresholds of oxygen saturation. In that meta-analysis, the best cutoff in discriminating kids with and without pneumonia was less than or equal to 96%. At that threshold, the likelihood ratio of having pneumonia was 2.8, so that altered your pretest probability of having pneumonia more significantly than other thresholds.

Florin: That's interesting. I think a lot of us consider 96% to be a pretty high threshold. Did you see any changes when you lowered that threshold? Certainly, in the International Diseases Society of America (IDSA) guidelines, you see 90%, and you see 92% in some other guidelines around the world. Did you see any threshold effects at those lower oxygen saturations?

Neuman: Yes, we probably would have. The problem is that many of the studies that looked at pneumonia had very few children who were actually hypoxic at those levels. It limits your ability to really evaluate lower levels of hypoxia for the presence or absence of radiographic pneumonia.

Florin: Let's unpack a little bit of that second criteria that you mentioned: work of breathing. How is work of breathing defined in these studies and in your meta-analysis?

Neuman: It was defined quite differently in all the different studies that looked at work of breathing. For the meta-analysis, we included any element of work of breathing: grunting, flaring, retracting. Any element of work of breathing was included in that combined variable for the purposes of a meta-analysis. Unfortunately, meta-analyses have to rely on the data in the form that it's collected. Because there's not a uniform nature in which these are assessed in different studies, we had to rely on however it was described in those individual studies.

Florin: That last symptom that you mentioned — wheezing — is negatively associated with radiographic pneumonia. I know you and I have done work in this area. What does wheezing say about how we can predict radiographic pneumonia in a child who's wheezing, given that it is negatively associated with radiographic pneumonia?

Neuman: I think this is interesting. When I did my fellowship, we were traditionally taught that children with asthma and bronchiolitis have higher rates of pneumonia. I think it's challenging to look at wheezing as a potential predictor of pneumonia. The issue is that most children who are undergoing an evaluation for potential pneumonia have signs or symptoms of asthma, reactive airway disease, wheezing, and bronchiolitis. The rate of pneumonia is actually much lower in those children than other children who present with similar signs, like hypoxia. It's a tough finding to look at with respect to diagnosing pneumonia. Among kids who are wheezing, the same sort of signs or symptoms (such as high fever or profound hypoxia) may make them more likely to have pneumonia. But even in those kids, the rate of pneumonia is very low among wheezing children.

Florin: Great. We've talked up to this point about these individual signs and symptoms. No one individual sign or symptom is diagnostic of radiographic pneumonia. There's been work, largely led by you, over the last decade about developing clinical prediction rules or scores that can be used to increase the accuracy of predicting pneumonia diagnosis on x-ray. Can you talk about those clinical prediction rules?

Neuman: The biggest thing that folks need to know is that physicians tend to overestimate the likelihood of pneumonia, at least based on radiographic findings. In a study we did many years ago, we asked physicians, "How likely do you think it is that this child will have radiographic pneumonia?" Across the board, physicians overestimated the likelihood of the patient having radiographic pneumonia. There are two thresholds I like to talk about. Among those children where physicians thought there was more than a 50% likelihood of pneumonia, only about 35%-40% of those kids had any radiographic finding at all, and only about 20% of those kids had radiographic pneumonia. At the highest threshold — greater than 75% — only 50% of those kids had any radiographic findings at all, and only 30% had definite radiographic pneumonia. I think the two factors that call for decision rules in this area are: (1) Physicians overestimate the likelihood of pneumonia, and (2) No individual finding performed well enough to hang your hat on. With that said, multiple groups have tried to develop clinical decision rules to help physicians better determine the need for either x-ray or antibiotic use in this group. The findings in a lot of these clinical decision rules showed that the individual factors are actually quite similar — things like older age, fever, fever duration, and focal lung findings like decreased breath sounds and rales. Those are findings that are typically incorporated in decision rules. Wheezing is also incorporated in many of the decision rules as a negative predictor of pneumonia. I think the interesting thing with these decision rules is that we found across the board, these rules perform much better than clinical judgment for the identification of radiographic pneumonia.

Florin: That's a great summary. This brings us to the next topic of discussion. I think there's still more work to be done to externally and widely validate these clinical prediction rules or scores. Let's say once validated, you apply some of these factors and have to make a decision. Do you treat this child empirically based on this clinical prediction rule, or this set of signs and symptoms? Or do you proceed with chest radiography in the diagnosis of pneumonia? How can we best use chest radiography in the diagnosis of pneumonia in children?

Neuman: I'll start by saying the IDSA recommends against the routine performance of chest x-ray in the outpatient setting when pneumonia is suspected. There are multiple reasons for that. The reasons cited in the guidelines include radiation exposure and the fact that chest x-ray can't reliably distinguish bacterial from viral infection. There's poor interrater reliability around specific radiographic findings. The most important one is that chest x-ray may not be accessible in all settings. The argument for chest x-ray, in my mind, relates to a couple of things. One is that clinical suspicion is not synonymous with radiographic pneumonia. Physicians tend to overestimate the likelihood of pneumonia to quite a big extent, and clinical findings are neither sensitive nor reliable for the prediction of radiographic pneumonia. One example of the argument for chest x-ray has to do with the fact that chest x-ray has a very high negative predictive value. In a study we conducted in our emergency department among 400 children in whom the clinician suspected radiographic pneumonia but had a normal or negative chest x-ray, only five children were subsequently diagnosed within the 2-week period following the emergency department visit. Thus, the negative predictive value of chest x-ray is 98.8%. Although a chest x-ray may not be particularly sensitive for the diagnosis of pneumonia, the negative predictive value is quite high.

Florin: Great. That's helpful to know. Another point that you made, which is one that I think is important to talk about, is this idea that any consolidation that you see on radiograph must be bacterial. The evidence suggests and the IDSA/Pediatric Infectious Diseases Society (PIDS) guidelines state that you cannot really differentiate viral from bacterial infection using a chest radiograph. Can you expand a bit on that point?

Neuman: I think there are certain radiographic features that make it more likely to be bacterial than viral, such as having pleural effusion or a large lobar consolidation. I think in those cases, most folks will suspect a bacterial etiology rather than a viral etiology. But most children with pneumonia actually have more subtle findings. Using radiographs to make the diagnosis of pneumonia has the challenge that most of the x-rays have findings that are not as likely for pneumonia.

Florin: Another challenge with the x-ray is that it can be notoriously difficult to interpret. We've all come across those x-rays where you see something, and you're not exactly certain if it's consolidation, atelectasis, or just peribronchial thickening that's masquerading as one of those findings. You mentioned the interrater reliability of a chest x-ray. We know that for a good reference standard, we really have to have a really reliable reference standard. Can you talk a bit about the limitations of x-ray in terms of its interrater reliability and specifically the reliability of different findings on x-ray?

Neuman: Sure. We studied radiologists and their interpretation of x-ray and looked at different findings on x-ray to see how often radiologists agree upon certain findings. We observed that, overall, the agreement for x-ray interpretation for things like infiltrate were quite low, with kappas in the range of 0.4-0.5. However, for the findings that are most suggestive of a bacterial etiology, like a pleural effusion or a lobar infiltrate, the agreement was a little bit higher. Overall, there is a lot of variation in the interpretation of x-ray, but the agreement was quite a bit higher for certain findings that are most suggestive of a bacterial etiology.

Florin: You alluded to obtaining a chest x-ray when the diagnosis might be uncertain — maybe those kids in intermediate risk — and avoiding x-ray when you're highly confident in the diagnosis or that the child does not have pneumonia. It's a little bit of a balance, particularly in the outpatient setting, where you may not have radiography immediately available to you. Do you obtain the chest radiograph and avoid potential antibiotics vs. empirically prescribing antibiotics without a chest radiograph and potentially overprescribing antibiotics? It's a bit of a tension, right? Can you discuss that tension between radiograph use and antibiotic use?

Neuman: In the outpatient primary care setting, where most children with suspected pneumonia are being treated and x-ray may not be as accessible, I think clinical decision rules can play a big role. Certainly, children who have a high level of suspicion of pneumonia probably don't need an x- ray as long as you're not concerned about a complicated type of pneumonia or pleural effusion. At the low end of the spectrum — children in whom the suspicion is quite low — I think those kids also don't need x-rays. The majority of children live in this zone of intermediate risk. For those kids, I think chest x-ray can be valuable in terms of reducing antibiotic use and potentially not treating children who likely do not have a radiographic finding.

Florin: An interesting part about all of this is the collateral damage of getting the x-ray. You alluded to the fact that most kids who have findings on an x-ray will have findings that are not going to be highly consistent with a consolidative pneumonia. There will be other findings. What's the collateral damage of getting that x-ray and seeing those other findings? The collateral damage is more antibiotic use, misdiagnosis, cost, and unnecessary radiation exposure. I completely agree with you that it really is a balance. The decision to get an x-ray is not always an insignificant one.

This was a great conversation. Today, I talked with Dr Mark Neuman about the diagnosis of pediatric pneumonia. A few takeaways I have from this conversation are that the diagnosis of pneumonia is challenging, and there is substantial overlap in the clinical symptoms of pediatric pneumonia with other common respiratory conditions, including bronchiolitis and asthma. We heard that no single sign or symptom is diagnostic of radiographic pneumonia, and we can use combinations of signs and symptoms in clinical prediction rules to help hone that accuracy to better predict which kids will have pneumonia on radiograph. The findings that are most important in combination are things like older age, longer fever duration, focal lung findings, and hypoxemia. Certainly, we think about wheezing as a negative predictive factor for having pneumonia on radiograph. The chest x-ray may be useful to exclude a clinically suspected pneumonia or to identify a complicated pneumonia, but routine chest x-ray is generally not recommended for well-appearing kids with suspected pneumonia who can be managed in the outpatient setting. The use of clinical scoring systems or clinical prediction models may help to reduce x-ray utilization and promote judicious use of antibiotics for children with respiratory illness. With that, I want to thank you for tuning in. If you haven't done so already, please take a moment to download the Medscape app to listen and subscribe to this podcast series on pediatric pneumonia. This is Dr Todd Florin for InDiscussion.


Pediatric Pneumonia

Boston Febrile Infant Algorithm 2.0: Improving Care of the Febrile Infant 1-2 Months of Age

Imaging in Pediatric Pneumonia

Does This Child Have Pneumonia?: The Rational Clinical Examination Systematic Review

The Management of Community-acquired Pneumonia in Infants and Children Older Than 3 Months of Age: Clinical Practice Guidelines by the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America

Physician Assessment of the Likelihood of Pneumonia in a Pediatric Emergency Department

Predicting Pneumonia From the Clinical Exam

Negative Chest Radiography and Risk of Pneumonia

Variability in the Interpretation of Chest Radiographs for the Diagnosis of Pneumonia in Children

Interrater Reliability: The Kappa Statistic

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.