Hot Off the Press: Accuracy of Signs and Symptoms for the Diagnosis of Community‐acquired Pneumonia

Community-acquired pneumonia (CAP) is a significant source of morbidity and mortality in adults.1, 2 However, most patients presenting with symptoms of acute lower respiratory tract infection (LRTI) will not have CAP. As a result, antibiotic overuse is well documented, potentially increasing rates of antimicrobial resistance as well as increasing costs of care.3, 4 However, ordering a chest x-ray on every patient with LRTI symptoms likely leads to unnecessary harm and costs. Therefore, it is important to know what features of the history and physical examination modify the risk of CAP, so that imaging and antibiotic prescriptions can be appropriately selected. These authors performed a systematic review and meta-analysis examining the accuracy of signs and symptoms in the diagnosis of CAP.5 This is a systematic review and meta-analysis examining the accuracy of signs and symptoms in diagnosis CAP in the outpatient setting. There were 16 studies that fit their inclusion criteria, with a range of 52 to 2,850 patients included. Chest x-ray was used as the criterion standard in all studies. The prevalence of CAP was 10% in primary care settings and 20% in the emergency department (ED). No individual sign or symptom was good enough to either rule in or rule out CAP. The most helpful indicator was “overall clinical impression,” with a positive likelihood ratio of 6.32 (95% confidence interval [CI] = 3.58 to 10.5) and a negative likelihood ratio of 0.54 (95% CI = 0.46 to 0.64). When assessing the quality of a systematic review, there are two major factors to consider: the quality of the search and quality of the studies that were found. This meta-analysis was registered with the PROSPERO database and followed the PRISMA guidelines for performing a systematic review. The quality of the search is good, although only a single database was searched, so it is possible that some studies may have been missed if they were not indexed in Medline. Furthermore, patients were excluded if they were from skilled nursing facilities, had chronic lung disease, or were immunosuppressed, so the results might not extrapolate to those populations. There are a number of potential sources of bias in the underlying studies that could influence the validity of the results. They report a 20% prevalence of CAP in the ED, but that seems quite high to us and could represent selection bias. Although we are generally taught that sensitivity and specificity are independent of disease prevalence, that is not true if the severity of disease also changes (known as spectrum bias). Presumably, if selection bias did occur and the prevalence is higher than expected, the patients included probably have more severe disease than those left out, and therefore, the sensitivity and negative predictive values might look better if they were assessed in all comers. Another potential source of bias to consider is the imperfect criterion standard. Chest x-ray is far from perfect when it comes to diagnosing CAP, with false positives, false negatives, and significant inconsistency in interpretation.6, 7 The results reported here assume that the chest x-ray was correct, which is how we generally practice clinically, but should limit our confidence in the reported numbers. Despite the limitations of these data, we think it would be a mistake to be too nihilistic when assessing these numbers. Although no individual sign or symptom was independently good enough to rule in or rule out CAP, the clinician’s overall impression was moderately accurate, and that impression was presumably based on a combination of these individual signs and symptoms. Therefore, despite the underwhelming numbers, it would be a mistake to interpret these results as indicating that the physical examination is futile or should be abandoned. The authors include 16 studies, which encompass a total of 8,507 patients. The prevalence of CAP was 20% in the ED studies and 10% in the primary care setting. No individual sign or symptoms 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 to 10.5) and a negative likelihood ratio of 0.54 (95% CI = 0.46 to 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. The finding with the best test characteristic to rule in pneumonia was egophony, with a positive likelihood ratio of 6.17 (95% CI = 1.34 to 18.0) when present, although the negative likelihood ratio was only 0.96 (95% CI = 0.93 to 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 to 0.48). These results tell us that, although we cannot rely on any individual sign or symptom for the diagnosis of CAP, physicians should be relatively comfortable trusting their overall clinical impression. The reported numbers may be helpful in teaching and refining doctors’ clinical judgment. ags (@Ags_win): Clinical diagnosis based on hx and PE findings however PE does not supersede hx. Lifelong Seattle Kraken fan (@movinmeat) responds: Hot take: the only useful element of the physical exam for CAP is the doorway eyeball exam. (Color, attentiveness, work of breathing, and ok vital signs). The stethoscope is a relic of the 20th century and may be safely left in your bag. Justin Hensley, MD FAWM (@EBMgoneWILD) responds: COVID has allowed me just that. No longer do I carry the stethoscope into patient rooms. Seth Trueger (@MDaware) responds: that any one element isn’t dispositive doesn’t mean the exam isn’t helpful; elements are useful together, not in isolation. Ryan Radecki, MD MS (@emlitofnote) responds:… as if “pneumonia” is a homogeneous presentation, regardless of causative etiology and host factors. Fabian Juzek (@mfkuepp) responds: Also I think we should be more "brave", especially if a patient is going to be admitted anyway, and withhold ABx if there's diagnostic uncertainty. Often there's no gold standard and we're running in circles. Then RCTs w treatment based on different diagnostic criteria should be done IMO. Maarten Van Hemelen (@Elennaro) responds: There's rather few diseases that have a good practical gold standard! For pna we at least have biopsy/autopsy. Obv not practical in all comers but esp autopsy is badly underused IMHO. We should be calibrating our decision-making, esp (but not excl) when outcomes are bad! Maarten Van Hemelen (@MaartenVHemelen): With some exceptions, I think most people should be irradiated (as you state it) before getting antibiotics. The kids of a cxr radiation dose are, in my mind, far lower than the harms of non-indicated abx. This goes a fortiori in previously treated patients. Dr. Ken Milne (@TheSGEM) responds: As an ID and critical care doc your population is probably different than the ED patient. Guidelines recommend against routine CXR to Dx CAP in kids. I don't treat children so couldn't comment on that, risks likely to be higher for ionizing radiation I guess? In adults I think POCUS would be OK or better. Problem with forgoing imaging entire. NB I recognize that my stance on this isn't generalisable to all situations and I've actually had some animated discussions about this with my primary care friends! Bottom line IMHO is that you should somehow acknowledge that most pts with rti don't need abx. Often ignored!ly, at least in Belgium, is that you get a lot of abx for bronchitis. Casey Parker (@broomedocs): You know the answer… Ultrasound … it is always the answer! No individual sign or symptoms is good enough to rule in or rule out CAP. Physicians should rely on their clinical judgment to determine which patients with LRTI symptoms require imaging or treatment for CAP.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.