Evidence-Based Disposition of Community-Acquired Pneumonia

Authors: Trevor Cerbini, MD (EM Resident Physician, Kings County Hospital Center/University Hospital Brooklyn) and Richard Sinert, DO (EM Attending Physician, Kings County Hospital Center/University Hospital Brooklyn) // Reviewed by: Andrew Grock, MD (West LA VA Hospital System and David Geffen SOM at UCLA); Alex Koyfman, MD (@EMHighAK); Brit Long, MD (@long_brit)


On starting what you hope will be a fairly calm Sunday morning shift, you take your first, delicious sip of life-giving coffee. But before you can fully enjoy it, a new patient pops up in your queue. You set aside your coffee, review the triage note and go to see the patient – a 32-year-old female with a past medical history of celiac disease presenting with two days of fevers, measured at home to 101.2°F (38.4°C), fatigue, and a mildly productive cough. She denies any recent illnesses or antibiotic use within the last few months. The patient had taken a combination cold & flu remedy 2 hours prior to arrival. You note on triage vitals that the patient is slightly tachycardic to 103 b/min, is otherwise normotensive, afebrile, and without tachypnea or hypoxia. A physical exam is notable for crackles in the left posterior and lateral lung fields. You order a two-view chest x-ray which reveals an obvious left lower lobar consolidation.

You head back to your workstation to take another sip of sweet, sweet coffee, and pat yourself on the back for nailing the diagnosis. Suddenly, your charge nurse grabs you by the scrub top and pulls you into the critical care area of the ER. “He doesn’t look so good, doc” is all your charge nurse tells you before dropping you at the feet of a patient who, indeed, doesn’t look so good at all. This 68-year-old man has a past medical history of insulin-dependent diabetes, hypertension, coronary artery disease with prior stents, on aspirin, and had been doing fairly well up until about 3 days ago when he developed generalized malaise, subjective fevers, worsening shortness of breath on exertion, and a harsh cough productive of mostly whitish but occasionally rusty sputum. The patient also endorses right-sided chest pain that worsens with deep inspiration and coughing. The vital signs are: 101.6 F (38.7 °C), HR 119, BP 140/75 mmHg, RR 24, and SaO2 96% on room air. The patient appears elderly, chronically ill and in no acute distress. Though he has normal mental status, he is slightly diaphoretic with skin that is warm to touch, and has rhonchi and crackles in the right anterior lung fields. A portable chest x-ray reveals a faint consolidation in the right middle lobe. A fingerstick glucose is 220 mg/dL. Abnormal labs include a leukocytosis of 12 x 109/L, a BUN of 35, and a creatinine of 1.7 (above the patient’s baseline of 1.2 on chart review). Otherwise, his complete blood count, comprehensive metabolic panel, and venous oximetry results are within normal limits. As you ponder the appropriate disposition for both patients, you feel a presence at your left and turn to see the imposing figure of your attending. She smirks at you and offers the simple question you had been dreading: “so what’re their dispos?”.


Hopefully the prior two cases illustrate the spectrum of disease with which Community Acquired-Pneumonia (CAP) can present. Given this variability, it can at times be difficult to determine the appropriate disposition for our patients. In this post we will review the evidence behind determining appropriate dispositions in CAP and hope to provide a clear framework for making these decisions. For diagnosis and management of CAP (and a wee bit on disposition), we refer the reader to a recent comprehensive emDOCs blog post by Anton Helman.

In October 2019 the Infectious Disease Society of America (IDSA) and the American Thoracic Society (ATS) published joint updated guidelines for the diagnosis and treatment of CAP, representing the first such update since the guidelines’ initial publication in 20071. These guidelines are supported by the American College of Emergency Physicians’ (ACEP) 2021 clinical policy on the management of adults with CAP2. In these updated guidelines, the authors make two important recommendations regarding the disposition of adult patients with CAP that are worth reviewing in detail.

First, the IDSA/ATS guidelines recommend that when determining the appropriateness of either inpatient versus outpatient management a validated clinical prediction rule for prognosis be used in addition to clinical judgment, rather than clinical judgment alone. Of the two clinical prediction rules,  they recommend the Pneumonia Severity Index (PSI) score over the CURB-65 score, as the PSI score identifies a larger proportion of low-risk patients and has a higher discriminative power in predicting 30-day mortality1,3.

The PSI score stratifies patients according to 30-day mortality into 5 different “risk classes”. Groups I and II are considered low risk (with 30-day aggregate mortality rates of ~ 1% depending on the patient population studied) and suitable for outpatient treatment (barring other considerations; keep reading!), while group III is deemed intermediate risk (suitable for possible short observation stay vs closely monitored outpatient treatment), and groups IV and V are considered moderate and high risk, respectively, requiring inpatient care.

The PSI score can be applied in a two-step process. First, an abbreviated screen can be done as follows:

– Age > 50 years

– Age < 50 years but with a hx of neoplastic disease, congestive heart failure, cerebrovascular disease, renal or liver disease

– Altered mental status, HR ≥ 125, RR ≥ 30, SBP ≤ 90, Temp < 35°C or ≥ 40°C


If all of the above are absent, the patient is considered risk class I; otherwise, the full 20 variable PSI score should be calculated (See Figure 1).

There are some (arguably minor) barriers to using the PSI score in clinical practice. The full PSI score is relatively complicated, consisting of 20 individual variables, however, the widespread availability of clinical calculators (especially when incorporated into the electronic medical record itself) makes this mostly irrelevant. In addition, 7 of these variables are based on laboratory and radiographic data, necessitating (at the minimum) a basic metabolic panel, complete blood count, venous blood gas, and chest radiograph be obtained.

If strictly adhered to an arterial blood gas (ABG) would also be required given that the partial pressure of O2 < 60mm Hg (PaO2)  is another laboratory parameter. We believe a solid argument could be made against obtaining an ABG in most patients as an oxygen saturation of at or above 91% with good waveform on pulse oximetry is highly suggestive of a correspondingly high PaO2 (above 60 mm Hg). This is supported by several investigations. First, Zeseron et al found good correlation between SpO2 and PaO2 in undifferentiated critically ill patients, with 96.4% of patients with SpO2 ≥ 90% having an SaO2 ≥ 60 mm Hg on ABG3. Of the 7 patients with lower PaO2 than expected, 5 were alkalemic, a factor known to left shift the oxygen-hemoglobin dissociation curve3. Thus, when using SpO2 to approximate PaO2, it is crucial to keep in mind the several factors that may left shift the oxygen-hemoglobin dissociation curve (eg, alkalemia, hypocarbia, hypothermia, CO poisoning, etc.). Other investigators have used clinical calculators requiring PaO2 (eg, SOFA score, P:F ratio) utilizing the Severinghaus equation in both critically ill and acute respiratory distress syndrome (ARDS) patients, suggesting again the largely predictable relationship between these two parameters4,5. Finally, the PaO2 only provides 10 points towards the score. Thus, a PSI without an ABG can at most be off by 10 points. We refer the reader to an excellent blog post by Josh Farkas at PulmCrit reviewing the (dis)utility of ABGs.

Of course, patients are not simply lists of variables, and many other factors should be considered when determining disposition. Additionally, even validated clinical prediction tools have limitations that must be considered. For example, we should not apply PSI to populations excluded from its derivation and validation studies, such as patients who are: pregnant, immunosuppressed, hospitalized within the previous 10 days, and with limited ability to follow outpatient instructions due to psychosocial factors6,7. An additional limitation is the inability to account for the severity of individual pneumonia risk factors. Hypoxia for example, gives you 10 points on the PSI for a PaO2 < 60 mm Hg, but no one would discharge an otherwise healthy 30-year-old with an SaO2 of 40% on room air just because they have a class II PSI score. Similarly, if a patient has a history of infections with multi-drug resistant organisms without an effective oral antibiotic option, he or she would seem to require admission for intravenous antibiotics regardless of the PSI score.

Highlighting the above limitations, discordance between PSI score-based disposition and actual disposition in clinical practice is well documented, and several studies (both prospective and retrospective) reflect that anywhere from 30-60% of disposition decisions are “discordant” for patients deemed low risk by PSI score7-9. Taken together, the results of these studies highlight a number of factors that influence the decision by providers to admit patients deemed “low risk” by PSI score; significant hypoxia/hypoxemia, both stable and unstable comorbidities, failure of initial outpatient therapy, signs of clinical deterioration, poor baseline functional status, and psychosocial factors are all among the most common7-9. In Marcos et al, signs of clinical deterioration included findings such as high fever, thoracic pain, hypotension, tachycardia, tachypnea, poor general condition, and abnormal ascultaton9. While many of these seem reasonable, in their prospective study of admitted CAP patients with low PSI scores, Marcos et al9 found that the only factors statistically significantly associated with adverse 30-day outcomes among the many cited by providers for their disposition decision were hypotension, tachycardia, suspicion of sepsis, and renal failure.

Without plunging too much further down the rabbit hole (too late?), the preceding discussion illustrates the importance of being a reasonable human-being clinician and not a cold unfeeling cyborg in our application of the PSI score. We should use the PSI (or validated clinical prediction tool of your choice) in conjunction with our clinician judgment, and keep in mind the exclusion criteria and its limitations.

While the need for intensive care unit (ICU) level care becomes obvious in patients with frank hypotension requiring vasopressors or with respiratory failure requiring mechanical ventilation, in other cases deciding between admission to the ICU versus the floor can be difficult. The importance of making this decision from the ER is highlighted by a number of retrospective studies that display a significant increase in mortality among patients upgraded to the ICU after admission to a general medical ward vs. patients directly admitted to the ICU1,10. The IDSA/ATS guidelines and ACEP both support use of the 2007 IDSA/ATS Minor Criteria in conjunction with clinician judgment to determine level of care. The minor criteria consist of 9 variables (see Table 1). In a 2011 meta-analysis, the presence of three minor criteria displayed a pooled sensitivity of 56% and a specificity of 91% for predicting ICU admission, with a positive likelihood ratio of 6.2211.

Case Resolutions:

Your first patient, the 32-year-old female, has none of the characteristics in Step 1 of the PSI score (see Figure 1 above and accompanying text), and is considered risk class I. The patient voices understanding of your home care instructions and return precautions, assures you that she will pick up the script for amoxicillin 1 gram TID for 5 days you sent to her pharmacy (which you prescribed after having reviewed the rest of the latest IDSA/ATS guidelines regarding CAP management; thanks Anton!), and agrees to check in with her primary care provider in the next 2-3 days.

Determining disposition for your second patient, the 68-year-old male, requires calculation of the full PSI score given his older age (not to mention his presentation overall). Based on his age and elevated BUN alone, his PSI score is 88, placing him in Risk class III. While technically a candidate for inpatient vs. OBS or even closely monitored outpatient management based on his risk class, being an astute clinician (and not a clinical prediction tool applying android), you also realize that this patient meets sepsis criteria given his elevated creatinine as a marker for end organ dysfunction. Along with ordering intravenous fluids, antipyretics, and ceftriaxone and azithromycin IV, you discuss the plan with your patient, who reluctantly agrees with admission. During a short ED observation, he maintains his SaO2 above 94% on room air, displays no increasing work of breathing or respiratory distress, and defervesces with normalization of his heart rate. Now reassured that he does not require ICU level care based on your repeat clinical assessments and the presence of only 1 IDSA/ATS 2017 minor criterion, you admit him to the floor.

You return to your workstation, and while typing out your notes reach out absentmindedly to where your coffee cup once stood, only to find nothingness; looking down at your desk, you see a crescent moon stain of sweet black elixir, the only proof of its existence. Your head nurse, seeing you crumpling in on yourself like a dying star, places your cup back in front of you and chides you, “Joint Commission could be here any minute. Next time I’m throwing it out!”. You laugh a nervous laugh.. and clutch your precious ever tighter.

Take Home Points:

  • ACEP and the IDSA/ATS guidelines recommend a clinical prediction rule in combination with clinician judgement to determine patient disposition for adults with CAP.
  • The PSI score identifies a large portion of low-risk patients potentially suitable for outpatient care, but should be used with clinical judgment! Be aware of excluded populations that the PSI score should not be used in.
  • Consider admitting patients with significant hypoxia, hypotension, tachycardia, or renal failure, regardless of PSI score.
  • Likewise, patients demonstrating an inability to comply with an outpatient antibiotic regimen (psychosocial factors, substance use, cognitive deficiencies, or PO intolerance) should also be considered for inpatient care.
  • ED admission to a higher-level care setting should be considered for patients with 3 or more IDSA/ATS 2007 minor criteria.

References/Further Reading: 

  1. Metlay J et al. Diagnosis and Treatment of Adults with Community-acquired Pneumonia. An Official Clinical Practice Guideline of the American Thoracic Society and Infectious Diseases Society of America. Am J Respir Crit Care Med. 2019;200:e45-e67.
  2. Smith et al. Clinical Policy: Critical Issues in the Management of Adult Patients Presenting to the Emergency Department With Community-Acquired Pneumonia. Ann Emerg Med. 2021;77:e1-e57.
  3. Zeseron E, Goodgame B et al. Correlation of Venous Blood Gas and Pulse Oximetry With Arterial Blood Gas in the Undifferentiated Critically Ill Patient. J Intensive Care Med. 2018;33:176-181.
  4. Brown SM, Grissom CK et al. Nonlinear Imputation of Pao2/Fio2From Spo2/Fio2 Among Patients With Acute Respiratory Distress Syndrome. 2016;150:307-313
  5. Pandharipande PP, Shintani AK et al. Derivation and validation of SpO2/FiO2ratio to impute for PaO2/FiO2 ratio in the respiratory component of the Sequential Organ Failure Assessment (SOFA) Score. Crit Care Med. 2009;37:1317-1321.
  6. Aujesky D et al. Prospective comparison of three validated prediction rules for prognosis in community-acquired pneumonia. Am J Med. 2005;118:384-392. https://doi.org/10.1016/j.amjmed.2005.01.006.
  7. Seymann G et al. Clinical judgment versus the pneumonia severity index in making the admission decision. J of Emerg Med. 2008; 34:261–268.
  8. Aujesky D et al. Reasons why emergency department providers do not rely on the pneumonia severity index to determine the initial site of treatment for patients with pneumonia. Clin Infect Dis. 2009;49:e100-e108.
  9. Marcos PJ et al. Discordance of physician clinical judgment vs. pneumonia severity index (PSI) score to admit patients with low-risk community-acquired pneumonia: a prospective multicenter study. J Thorac Dis. 2017; 9:1538-1546.
  10. Restrepo MI, Mortensen EM, Rello J, Brody J, Anzueto A. Late admission to the ICU in patients with community-acquired pneumonia is associated with higher mortality. 2010;137:552–557.
  11. Chalmers JD, Mandal P, Singanayagam A, Akram AR, Choudhury G, Short PM, et al. Severity assessment tools to guide ICU admission in community-acquired pneumonia: systematic review and meta-analysis. Intensive Care Med. 2011;37:1409–1420.

Leave a Reply

Your email address will not be published. Required fields are marked *