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

Overview
In the more than 10 years since the last American Thoracic Society(ATS)/Infectious Diseases
Society of America (IDSA) community-acquired pneumonia (CAP) guideline (1), there have been
changes in the process for guideline development, as well as generation of new clinical data.
ATS and IDSA agreed on moving from the narrative style of previous documents to the Grading
of Recommendations Assessment, Development, and Evaluation (GRADE) format. We thus
developed this updated CAP guideline as a series of questions answered from available
evidence in an “is option A better than option B” format using the Patient or Population,
Intervention, Comparison, Outcome (PICO) framework (2).
Given the expansion in information related to the diagnostic, therapeutic, and management
decisions for the care of patients with CAP, we have purposely narrowed the scope of this
guideline to address decisions from the time of clinical diagnosis of pneumonia (i.e., signs and
symptoms of pneumonia with radiographic confirmation) to completion of antimicrobial
therapy and follow-up chest imaging. The document does not address either the initial clinical
diagnostic criteria or prevention of pneumonia. CAP is an extraordinarily heterogeneous illness,
both in the range of responsible pathogens and the host response. Thus, the PICO questions we
identified for this guideline do not represent the full range of relevant questions about the
management of CAP but encompass a set of core questions identified as high priority by the
panel. In addition, although each question was addressed using systematic reviews of available
high-quality studies, the evidence base was often insufficient, emphasizing the continued
importance of clinical judgment and experience in treating patients with this illness and the
need for continued research.

Introduction
This guideline addresses the clinical entity of pneumonia that is acquired outside of the hospital
setting. Although we recognize that CAP is frequently diagnosed without the use of a chest
radiograph, especially in the ambulatory setting, we have focused on studies that used
radiographic criteria for defining CAP, given the known inaccuracy of clinical signs and
symptoms alone for CAP diagnosis (3). This guideline focuses on patients in the United States
who have not recently completed foreign travel, especially to regions with emerging respiratory
pathogens. This guideline also focuses on adults who do not have an immunocompromising
condition, such as inherited or acquired immune deficiency or drug-induced neutropenia,
including patients actively receiving cancer chemotherapy, patients infected with HIV with
suppressed CD4 counts, and solid organ or bone marrow transplant recipients. Antibiotic
recommendations for the empiric treatment of CAP are based on selecting agents effective
against the major treatable bacterial causes of CAP.
Traditionally, these bacterial pathogens include Streptococcus pneumoniae, Haemophilus
influenzae, Mycoplasma pneumoniae, Staphylococcus aureus, Legionella species, Chlamydia
pneumoniae, and Moraxella catarrhalis. The microbial etiology of CAP is changing, particularly
with the widespread introduction of the pneumococcal conjugate vaccine, and there is
increased recognition of the role of viral pathogens. The online supplement contains a more
detailed discussion of CAP microbiology. As bacterial pathogens often coexist with viruses and
there is no current diagnostic test accurate enough or fast enough to determine that CAP is due
solely to a virus at the time of presentation (see below), our recommendations are to initially
treat empirically for possible bacterial infection or coinfection. In addition, the emergence of
multidrugresistant pathogens, including methicillinresistant S. aureus (MRSA) and
Pseudomonas aeruginosa, requires separate recommendations when the risk of each of these
pathogens is elevated. We acknowledge that other multidrug-resistant Enterobacteriaceae can
cause CAP, including organisms producing extendedspectrum b-lactamase, but we do not
discuss them separately because they are much less common and are effectively covered by the
strategies presented for P. aeruginosa. Therefore, throughout this document when discussing P.
aeruginosa we are also referring to other similar multiresistant gram-negative bacteria.
We have maintained the convention of separate recommendations on the basis of the severity
of illness. Although historically site of care (outpatient, inpatient general ward, or ICU) has
served as a severity surrogate, decisions about site of care may be based on considerations
other than severity and can vary widely between hospitals and practice sites.
We have therefore chosen to use the IDSA/ATS CAP severity criteria that have been validated
and define severe CAP as present in patients with either one major criterion or three or more
minor criteria. (Table 1)
This guideline reaffirms many recommendations from the 2007 statement. However, new
evidence and a new process have led to significant changes, which are summarized in Table 2.
Methods
The guideline development methodology and how conflict of interest was managed are
presented in the online supplement. In brief, the list of PICO questions was finalized based on a
prioritization of the most important management decisions balanced against the decision to
reduce the overall length of the document and total number of recommendations to maximize
readability and usability. We followed the GRADE standards for evaluating the evidence for
each PICO and assigned a quality of evidence rating of high, moderate, low, or very low. On the
basis of the quality of evidence, recommendations were assigned as strong or conditional. In
some cases, strong recommendations were made in the setting of low or very low quality of
evidence in accordance with the GRADE rules for when such recommendations are allowable
(e.g., when the consequences of the recommendation were high, such as preventing harm or
saving life). In all other cases, recommendations that were based on low or very low quality of
evidence and not believed to represent standards of care were labeled as conditional
recommendations.
Statements in favor of strong recommendations begin with the words “We recommend . . .”;
statements in favor of conditional recommendations begin with the words “We suggest . . . .”
Although we specified pairwise PICO questions for all antibiotic options in the outpatient and
inpatient settings, we summarized the recommendations using lists of treatment options, in no
preferred order, rather than retain the PICO format for this section.
Recommendations

Question 1: In Adults with CAP, Should Gram Stain and Culture of Lower Respiratory Secretions
Be Obtained at the Time of Diagnosis?
Recommendation. We recommend not obtaining sputum Gram stain and culture routinely in
adults with CAP managed in the outpatient setting (strong recommendation, very low quality of
evidence).
We recommend obtaining pretreatment Gram stain and culture of respiratory secretions in
adults with CAP managed in the hospital setting who:
1. are classified as severe CAP (see Table 1), especially if they are intubated (strong
recommendation, very low quality of evidence); or
2.
a. are being empirically treated for MRSA or P. aeruginosa (strong recommendation, very
low quality of evidence); or
b. were previously infected with MRSA or P. aeruginosa, especially those with prior
respiratory tract infection (conditional recommendation, very low quality of evidence);
or
c. were hospitalized and received parenteral antibiotics, whether during the
hospitalization event or not, in the last 90 days (conditional recommendation, very low
quality of evidence).
Summary of the evidence. Arguments for trying to determine the etiology of CAP are that 1) a
resistant pathogen may be identified; 2) therapy may be narrowed; 3) some pathogens, such as
Legionella, have public health implications; 4) therapy may be adjusted when patients fail initial
therapy; and 5) the constantly changing epidemiology of CAP requires ongoing evaluation.
These arguments stand in contrast to the lack of high-quality evidence demonstrating that
routine diagnostic testing improves individual patient outcomes. Studies that specifically
evaluated the use of sputum Gram stain and culture alone (4–7), or in combination with other
microbiological testing (8–11), also did not demonstrate better patient outcomes.
The overall poor yield of sputum evaluation for detecting organisms causing CAP limits its
impact on management and patient outcomes. Obtaining a valid sputum specimen can be
challenging because of patient-related characteristics (12–17). Performance characteristics of
testing also vary by organism, receipt of prior antibiotics, and setting. For example, in patients
with bacteremic pneumococcal pneumonia who have not received antibiotics, microscopic
examination and culture of a good-quality sputum sample detects pneumococci in 86% of cases
(18).
Rationale for the recommendation. In balancing the lack of evidence supporting routine sputum
culture with the desire for improved antimicrobial stewardship, the committee voted to
continue the stance of previous guidelines in recommending neither for nor against routinely
obtaining sputum Gram stain and culture in all adults with CAP managed in the hospital setting.
Whether to culture patients or not should be determined by individual clinicians on the basis of
clinical presentation, local etiological considerations, and local antimicrobial stewardship
processes.
The committee identified two situations in which we recommend sputum Gram stain and
culture: in hospitalized patients with severe CAP, and when strong risk factors for MRSA and P.
aeruginosa are identified, unless local etiological data have already shown these pathogens are
very infrequently identified in patients with CAP. Patients who have severe CAP requiring
intubation should have lower respiratory tract samples, such as endotracheal aspirates, sent for
Gram stain and culture promptly after intubation, particularly as these patients may be more
likely to have pneumonia due to MRSA or P. aeruginosa, and endotracheal aspirates have a
better yield of microbiological organisms than sputum culture (19).
We recommend obtaining sputum for Gram stain and culture in situations when risk factors for
MRSA or P. aeruginosa are present, both when initial empiric therapy is expanded to cover
these pathogens and when it is not expanded. In the former case, negative microbiological test
results may be used to deescalate therapy, and in the latter case, positive microbiological test
results may be used to adjust therapy. As discussed later, although there are numerous studies
identifying individual risk factors for MRSA and P. aeruginosa, many of these associations are
weak and vary across sites. The most consistently strong risk factor to consider is prior infection
with either MRSA or P. aeruginosa. In addition, hospitalization and treatment with parenteral
antibiotics in the last 90 days is associated with an increased risk of these pathogens, and so we
recommend sputum culture in this situation. These recommendations are not based on
highgrade evidence but reflect the committee’s desire to improve antibiotic use as well as
improve clinicians’ understanding of their local pathogen prevalences and resistance patterns,
which we believe are key to selecting appropriate empiric antibiotic therapy.
Research needed in this area. Rapid, cost-effective, sensitive, and specific diagnostic tests to
identify organisms causing CAP have potential to improve routine care by supporting the use of
targeted therapy, especially when there are risk factors for antibiotic-resistant pathogens. All
new diagnostic tests should be assessed in high-quality research studies that address the
impact of testing strategies on treatment decisions and patient outcomes.
Question 2: In Adults with CAP, Should Blood Cultures Be Obtained at the Time of Diagnosis?
Recommendation. We recommend not obtaining blood cultures in adults with CAP managed in
the outpatient setting (strong recommendation, very low quality of evidence).
We suggest not routinely obtaining blood cultures in adults with CAP managed in the hospital
setting (conditional recommendation, very low quality of evidence).
We recommend obtaining pretreatment blood cultures in adults with CAP managed in the
hospital setting who:
1. are classified as severe CAP (see Table 1) (strong recommendation, very low quality of
evidence); or
2.
 are being empirically treated for MRSA or P. aeruginosa (strong recommendation, very
low quality of evidence); or
 were previously infected with MRSA or P. aeruginosa, especially those with prior
respiratory tract infection (conditional recommendation, very low quality of evidence);
or
 were hospitalized and received parenteral antibiotics, whether during the
hospitalization event or not, in the last 90 days (conditional recommendation, very low
quality of evidence).
Summary of the evidence. There are no high-quality studies that specifically compared patient
outcomes with and without blood culture testing. One large observational study found lower
mortality for hospitalized patients associated with obtaining blood cultures at the time of
admission (20). Three subsequent (smaller) observational studies found similar associations
between in-hospital mortality and having blood cultures within 24 hours of admission, but the
results were not statistically significant (8, 21, 22).
The yield of blood cultures in most series of adults with nonsevere CAP is low, ranging from 2%
(outpatients) to 9% (inpatients) (14, 21, 23, 24); furthermore, blood cultures rarely result in an
appropriate change in empiric therapy (25), and blood specimens that include skin
contaminants can generate false-positive test results. Growth of organisms such as coagulase-
negative staphylococci, which are not recognized as CAP pathogens (26), may lead to
inappropriate antimicrobial use that increases the risk for adverse drug effects. A study of
adults hospitalized with CAP found blood cultures were associated with a significant increase in
length of stay and
duration of antibiotic therapy (27). Given the observational nature of these studies, it is
unknown whether the associations found with blood cultures and patient outcomes were
causal or due to unmeasured confounding factors, including severity of illness. Rationale for the
recommendation. Although additional diagnostic information could improve the quality of
treatment decisions, support for routine collection of blood cultures is reduced by the low
quality of studies demonstrating clinical benefit. Routinely obtaining blood cultures may
generate falsepositive results that lead to unnecessary antibiotic use and increased length of
stay.
In severe CAP, delay in covering lesscommon pathogens can have serious consequences.
Therefore, the potential benefit of blood cultures is much larger when results can be returned
within 24 to 48 hours.
The rationale for the recommendation for blood cultures in the setting of risk factors for MRSA
and P. aeruginosa is the same as for sputum culture.

Question 3: In Adults with CAP, Should Legionella and Pneumococcal Urinary Antigen Testing Be
Performed at the Time of Diagnosis?
Recommendation. We suggest not routinely testing urine for pneumococcal antigen in adults
with CAP (conditional recommendation, low quality of evidence), except in adults with severe
CAP (conditional recommendation, low quality of evidence).
We suggest not routinely testing urine for Legionella antigen in adults with CAP (conditional
recommendation, low quality of evidence), except
1. in cases where indicated by epidemiological factors, such as association with a
Legionella outbreak or recent travel (conditional recommendation, low quality of
evidence); or
2. in adults with severe CAP (see Table 1) (conditional recommendation, low quality of
evidence).
We suggest testing for Legionella urinary antigen and collecting lower respiratory tract
secretions for Legionella culture on selective media or Legionella nucleic acid amplification
testing in adults with severe CAP (conditional recommendation, low quality of evidence).
Summary of the evidence. Falguera and colleagues (28) randomized 177 patients to pathogen-
directed treatment (targeted treatment) on the basis of results of urinary antigen testing for S.
pneumoniae and Legionella versus empirical guidelinedirected treatment. Of the 88 patients in
the targeted treatment arm, 25% had a positive urinary antigen test and received
pathogendirected therapy. There were no statistical differences in death, clinical relapse, ICU
admission, length of hospitalization, or length of antibiotic treatment (28). A second trial of 262
patients included a broader range of microbiological testing (sputum and blood cultures) and
only Legionella urinary antigen testing, but patients receiving pathogen-directed therapy had
similar clinical outcomes to patients receiving empirical, guidelinedirected therapy, including
mortality, rates of clinical failure, and length of hospitalization (10).
One observational study evaluated cost and antibiotic selection in patients during two time
periods, with and without pneumococcal urinary antigen testing, but found no differences
during the two time periods (29). In contrast, other observational studies that have evaluated
the impact of prior CAP guideline concordance (including initial diagnostic testing with urinary
antigen tests and blood cultures, along with site of care stratification and guideline-concordant
therapy) have reported reduced mortality for patients receiving prior CAP guidelineconcordant
care, including diagnostic testing. Costantini and colleagues reported a 57% statistically
significant reduced odds of in-hospital mortality for patients receiving pneumococcal and
Legionella urinary antigen testing compared with patients not tested, adjusting for baseline
demographic and clinical differences (27). Uematsu and colleagues reported 25% reduced odds
of 30-day mortality in patients receiving urinary antigen tests but no impact on length of
hospitalizations (7). However, neither study distinguished whether the mortality benefits
attributed to testing were a direct consequence of the test results or a marker of other
improved processes of care.
Randomized trials have failed to identify a benefit for urinary antigen testing for S. pneumoniae
and Legionella. Concern has also been raised that narrowing therapy in response to positive
urinary antigen tests could lead to increased risk of clinical relapse (28). In large observational
studies, these diagnostic tests have been associated with reduction in mortality; therefore, we
recommend testing in patients with severe disease. An increase in Legionella infections in the
United States in the past decade highlights the importance of this diagnosis especially among
severely ill patients, particularly in the setting of potential outbreaks due to a common source,
although most cases are not associated with a known outbreak and remain sporadic (30, 31).
Research needed in this area. Newer nucleic acid amplification systems for sputum, urine, and
blood are being developed and require rigorous testing to assess the impact on treatment
decisions and clinical outcomes for patients with CAP, as well as the public health benefit in
terms of prevention of additional cases and informing primary prevention strategies. In
particular, we acknowledge the emergence of rapid, lowcost genomic sequence detection
assays that have the potential to greatly improve pathogen-directed therapy and thereby
improve antimicrobial stewardship.
Question 4: In Adults with CAP, Should a Respiratory Sample Be Tested for Influenza Virus at the
Time of Diagnosis?
Recommendation. When influenza viruses are circulating in the community, we recommend
testing for influenza with a rapid influenza molecular assay (i.e., influenza nucleic acid
amplification test), which is preferred over a rapid influenza diagnostic test (i.e., antigen test)
(strong recommendation, moderate quality of evidence).
Summary of the evidence. Rapid influenza tests have become increasingly available, moving
from earlier antigen-based detection tests to nucleic acid amplification tests. We were unable
to identify any studies that evaluated the impact of influenza testing on outcomes in adults with
CAP. In contrast, a substantial literature has evaluated the importance of influenza testing in
the general population, specifically among patients with influenza-like illness (32). Our
recommendations for influenza testing in adults with CAP are consistent with testing
recommendations for the broader population of adults with suspected influenza, as
summarized in the recent IDSA Influenza Clinical Practice Guideline
Rationale for the recommendation. The benefits of antiviral therapy support testing of patients
during periods of high influenza activity. During periods of low influenza activity, testing can be
considered but may not be routinely performed. Of note, this testing recommendation has both
therapeutic and infection-control implications in the hospital setting. Updated influenza testing
recommendations are also available on the CDC website
(https://www.cdc.gov/flu/professionals/ diagnosis/index.htm).

Question 5: In Adults with CAP, Should Serum Procalcitonin plus Clinical Judgment versus
Clinical Judgment Alone Be Used to Withhold Initiation of Antibiotic Treatment?
Recommendation. We recommend that empiric antibiotic therapy should be initiated in adults
with clinically suspected and radiographically confirmed CAP regardless of initial serum
procalcitonin level (strong recommendation, moderate quality of evidence).
Summary of the evidence. Several studies have assessed the ability of procalcitonin to
distinguish acute respiratory infections due to pneumonia (which are of viral or bacterial
etiology) from acute bronchitis or upper respiratory tract infections (which are almost
exclusively viral in etiology). However, for the purposes of this guideline, the question is
whether, among patients with clinically confirmed CAP, measurement of procalcitonin can
distinguish patients with viral versus bacterial etiologies and guide the need for initial antibiotic
therapy. Some investigators have suggested that procalcitonin levels of <0.1 mg/L indicate a
high likelihood of viral infection, whereas levels >0.25 mg/L indicate a high likelihood of
bacterial pneumonia (34–36). However, a recent study in hospitalized patients with CAP failed
to identify a procalcitonin threshold that discriminated between viral and bacterial pathogens,
although higher procalcitonin strongly correlated with increased probability of a bacterial
infection (37). The reported sensitivity of procalcitonin to detect bacterial infection ranges from
38% to 91%, underscoring that this test alone cannot be used to justify withholding antibiotics
from patients with CAP (38).
Rationale for the recommendation. Procalcitonin has been used to guide initiation of antibiotics
in patients with lower respiratory infections, but many of these studies are not restricted to
patients with radiographically confirmed pneumonia. Some patients with low procalcitonin
levels have CAP and have been safely treated without antibiotics (35), but these represent small
subgroups, raising concerns about the safety of widely using such a strategy.
Research needed in this area. Given the epidemiological evidence that viruses are an important
cause of CAP, there is a critical need to validate the use of current rapid laboratory tests,
including point-of-care tests, to accurately identify situations in which antibacterial therapy can
be safely withheld among adults with CAP.

Question 6: Should a Clinical Prediction Rule for Prognosis plus Clinical Judgment versus Clinical
Judgment Alone Be Used to Determine Inpatient versus Outpatient Treatment Location for
Adults with CAP?
Recommendation. In addition to clinical judgement, we recommend that clinicians use a
validated clinical prediction rule for prognosis, preferentially the Pneumonia Severity Index (PSI)
(strong recommendation, moderate quality of evidence) over the CURB-65 (tool based on
confusion, urea level, respiratory rate, blood pressure, and age >65) (conditional
recommendation, low quality of evidence), to determine the need for hospitalization in adults
diagnosed with CAP.
Summary of the evidence. Both the PSI and CURB-65 were developed as prognostic models in
immunocompetent patients with pneumonia, using patient demographic and clinical variables
from the time of diagnosis to predict 30-day mortality (39, 40). When compared with CURB-65,
PSI identifies larger proportions of patients as low risk and has a higher discriminative power in
predicting mortality (41). Two multicenter, cluster-randomized trials demonstrated that use of
the PSI safely increases the proportion of patients who can be treated in the outpatient setting
(42, 43). These trials and one additional randomized controlled trial (RCT) support the safety of
using the PSI to guide the initial site of treatment of patients without worsening mortality or
other clinically relevant outcomes (42–44). Consistent evidence from three pre–post
intervention studies and one prospective controlled observational study support the
effectiveness and safety of using the PSI to guide the initial site of treatment (45–48).
Clinical severity is not the only consideration in determining the need for hospital admission
(49, 50). Some patients have medical and/or psychosocial contraindications to outpatient
therapy, such as inability to maintain oral intake, history of substance abuse, cognitive
impairment, severe comorbid illnesses, and impaired functional status.
The PSI may underestimate illness severity among younger patients and oversimplify how
clinicians interpret continuous variables (e.g., all systolic blood pressures ,90 mm Hg are
considered abnormal, regardless of the patient’s baseline and actual measurement). Therefore,
when used as a decision aid, the PSI should be used in conjunction with clinical judgment.
In comparison to the PSI, there is less evidence that CURB-65 is effective as a decision aid in
guiding the initial site of treatment. One pre–post, controlled intervention study using an
electronically calculated version of CURB-65, PaO2/FIO2,300, absence of pleural effusion, and
fewer than three minor ATS severity criteria observed no significant increase in the use of
outpatient treatment for adults with CAP (51). A randomized trial compared the safety of
inpatient versus outpatient treatment of 49 patients with CURB-65 scores of less than 2 (52) but
had limited power to detect differences in patient outcomes; furthermore, outpatient
treatment included daily nursing visits and parenteral antibiotic therapy that is typically
restricted to inpatient care.

Rationale for the recommendation. Our recommendation to use the PSI as an adjunct to clinical
judgment to guide the initial site of treatment is based on consistent evidence of the
effectiveness and safety of this approach. Using a safe and effective decision aid to increase
outpatient treatment of patients with CAP has potential to decrease unnecessary variability in
admission rates, the high cost of inpatient pneumonia treatment (53, 54), and the risk of
hospital-acquired complications. Providing a conditional recommendation to use CURB-65
considers its greater simplicity of use relative to the PSI despite the paucity of evidence
regarding its effectiveness or safety.

Research needed in this area. It is important to study the effectiveness and safety of using CURB
scores or new prediction rules for prognosis as decision aids to guide the initial site of
treatment for patients with CAP compared with the PSI. Future studies of prediction rules
should also test electronic versions generated in real time from data routinely recorded in the
electronic medical record and assess their performance in patient populations excluded from
the development of existing prediction rules (55, 56).

Question 7: Should a Clinical Prediction Rule for Prognosis plus Clinical Judgment versus Clinical
Judgment Alone Be Used to Determine Inpatient General Medical versus Higher Levels of
Inpatient Treatment Intensity (ICU, Step- Down, or Telemetry Unit) for Adults with CAP?

Recommendation. We recommend direct admission to an ICU for patients with hypotension

For patients not requiring vasopressors or mechanical ventilator support, we suggest using the
IDSA/ATS 2007 minor severity criteria (Table 1) together with clinical judgment to guide the
need for higher levels of treatment intensity (conditional recommendation, low quality of
evidence).

Summary of the evidence. The PSI and CURB-65 were not designed to help select the level of
care needed by a patient who is hospitalized for CAP. Several prognostic models have been
designed to predict the need for higher levels of inpatient treatment intensity using severity-of-
illness parameters based on patient outcomes (ATS 2001, IDSA/ATS 2007, SMART-COP, and
SCAP score). Studies of prognostic models have used different end points, including inpatient
mortality (57, 58), ICU admission (57–59), receipt of intensive respiratory or vasopressor
support (59, 60), or ICU admission plus receipt of a critical therapy (61). In comparative studies,
these prognostic models yield higher overall accuracy than the PSI or CURB-65 when using
illness outcomes other than mortality (58, 59, 61).

The 2007 IDSA/ATS CAP guidelines recommended a set of two major and nine minor criteria to
define severe pneumonia requiring ICU admission (Table 1). These criteria were based on
empirical evidence from published studies and expert consensus. All elements are routinely
available in emergency department settings and are electronically calculable (51, 61). Several
groups have validated these criteria in pneumonia cohorts from different countries (57–59, 61),
with a meta-analysis reporting one major or three minor criteria had a pooled sensitivity of 84%
and a specificity of 78% for predicting ICU admission (62). Without the major criteria, a
threshold of three or more minor criteria (recommended in the 2007 IDSA/ATS guideline) had a
pooled sensitivity of 56% and specificity of 91% for predicting ICU admission (63).

SMART-COP is an alternative, validated prediction rule for identifying patients with pneumonia
who need vasopressor support and/or mechanical ventilation. The eight SMART-COP criteria
and the nine 2007 IDSA/ATS minor criteria have five overlapping elements: hypoxia, confusion,
respiratory rate, multilobar radiographic opacities, and low systolic blood pressure. SMART-COP
had a pooled sensitivity of 79% and specificity of 64% in predicting ICU admission using a
threshold of three or more criteria but uses albumin, PaO2, and pH, which are not universally
available for real-time clinical decisionmaking (60). For predicting ICU admission, one
comparison reported equivalence of the IDSA/ATS minor criteria and SMARTCOP (63) and
another reported a significantly greater performance of the IDSA/ATS minor criteria (61). No
randomized studies have evaluated the effectiveness or safety of an illness severity tool as a
decision aid to guide the intensity of inpatient treatment for patients hospitalized with CAP.

Rationale for the recommendation. Patients transferred to an ICU after admission to a hospital
ward experience higher mortality than those directly admitted to the ICU from an emergency
department (64–67). This higher mortality may in part be attributable to progressive
pneumonia, but “mis-triage” of patients with unrecognized severe pneumonia may be a
contributing factor (64). It seems unlikely that physician judgment alone would be equivalent to
physician judgment together with a severity tool to guide the site-of-care decision. We
recommend the 2007 IDSA/ATS severe CAP criteria over other published scores, because they
are composed of readily available severity parameters and are more accurate than the other
scores described above.
Research needed in this area. Controlled studies are needed to study the effectiveness and
safety of using illness severity tools as decision aids to guide the intensity of treatment in adults
hospitalized for pneumonia.

Question 8: In the Outpatient Setting, Which Antibiotics Are Recommended for Empiric
Treatment of CAP in Adults?
Recommendation.
1. For healthy outpatient adults without comorbidities listed below or risk factors for
antibiotic resistant pathogens, we recommend (Table 3):
 amoxicillin 1 g three times daily (strong recommendation, moderate quality of
evidence), or d doxycycline 100 mg twice daily (conditional recommendation, low
quality of evidence), or
 a macrolide (azithromycin 500 mg on first day then 250 mg daily or clarithromycin 500
mg twice daily or clarithromycin extended release 1,000 mg daily) only in areas with
pneumococcal resistance to macrolides ,25% (conditional recommendation, moderate
quality of evidence).
2. For outpatient adults with comorbidities such as chronic heart, lung, liver, or renal disease;
diabetes mellitus; alcoholism; malignancy; or asplenia we recommend (in no particular
order of preference) (Table 3):
 Combination therapy:
o amoxicillin/clavulanate 500 mg/125 mg three times daily, or amoxicillin/
clavulanate 875 mg/125 mg twice daily, or 2,000 mg/125 mg twice daily, or a
cephalosporin (cefpodoxime 200 mg twice daily or cefuroxime 500 mg twice
daily); AND
o macrolide (azithromycin 500 mg on first day then 250 mg daily, clarithromycin
[500 mg twice daily or extended release 1,000 mg once daily]) (strong
recommendation, moderate quality of evidence for combination therapy), or
doxycycline 100 mg twice daily (conditional recommendation, low quality of
evidence for combination therapy); OR
 Monotherapy:
o respiratory fluoroquinolone (levofloxacin 750 mg daily, moxifloxacin 400 mg
daily, or gemifloxacin 320 mg daily) (strong recommendation, moderate quality
of evidence).

Summary of the evidence. RCTs of antibiotic treatment regimens for adults with CAP provide
little evidence of either superiority or equivalence of one antibiotic regimen over another,
because of small numbers and the rare occurrence of important outcomes such as mortality or
treatment failure resulting in hospitalization. Several published trials included comparators that
are no longer available (e.g., ketolides). This paucity of data was noted in a 2014 Cochrane
review (68).

We identified 16 relevant RCTs comparing two antibiotic regimens for the treatment of
outpatient CAP (69–84). Metaanalyses of each of these groups of studies revealed no
differences in relevant outcomes between any of the compared regimens. Similar findings have
been reported in a 2008 meta-analysis of antibiotic treatment for outpatient CAP (85).

The committee also considered whether to accept data regarding oral antibiotics given to
inpatients with CAP. We believed that this evidence, albeit indirect, could be reasonably
extended to outpatients, because inpatients are generally higher risk and more severely ill. As
observational data suggest that inpatient and outpatient CAP are due to the same pathogens
(69, 71–73, 82), except for Legionella and gram-negative bacilli, which are rarely documented in
outpatient settings, it seems reasonable that an antibiotic regimen that was effective for
inpatients would be effective for outpatients.

Studies of high-dose oral amoxicillin have demonstrated efficacy for inpatients with CAP (86–
88). Similarly, there is evidence supporting amoxicillin-clavulanic acid in outpatient CAP (71, 73)
and inpatient CAP (89, 90).

There are limited data regarding oral doxycycline for pneumonia, mostly involving small
numbers of patients (81). Intravenous doxycycline 100 mg twice daily compared favorably to
intravenous levofloxacin 500 mg daily in 65 in patients with CAP (91). In an open-label
randomized trial of intravenous doxycycline 100 mg twice daily compared with standard
antibiotics, doxycycline was associated with a quicker response and less change in antibiotics
(92).

Rationale for the recommendation. Given the paucity of RCT data in the outpatient setting, the
committee considered all available evidence. The data included the few RCTs of outpatient CAP,
observational studies, RCTs of inpatient CAP treatment, antimicrobial resistance data from
surveillance programs, and data regarding antibiotic-related adverse events.

For patients without comorbidities that increase the risk for poor outcomes, the panel
recommended amoxicillin 1 g every 8 hours or doxycycline 100 mg twice daily. The
recommendation for amoxicillin was based on several studies that showed efficacy of this
regimen for inpatient CAP despite presumed lack of coverage of this antibiotic for atypical
organisms. This treatment also has a long track record of safety. The recommendation for
doxycycline was based on limited clinical trial data, but a broad spectrum of action, including
the most common relevant organisms. Some experts recommend that the first dose of oral
doxycycline be 200 mg, to achieve adequate serum levels more rapidly. There are no data
assessing whether such an approach is associated with improved outcomes.
In a departure from the prior CAP guidelines, the panel did not give a strong recommendation
for routine use of a macrolide antibiotic as monotherapy for outpatient CAP, even in patients
without comorbidities. This was based on studies of macrolide failures in patients with
macrolide-resistant S. pneumoniae (93, 94), in combination with a macrolide resistance rate
of .30% among S. pneumoniae isolates in the United States, most of which is highlevel
resistance (95). However, in settings where macrolide resistance is documented to be low and
there are contraindications to alternative therapies, a macrolide as monotherapy is a treatment
option.
Patients with comorbidities should receive broader-spectrum treatment for two reasons. First,
such patients are likely more vulnerable to poor outcomes if the initial empiric antibiotic
regimen is inadequate. Second, many such patients have risk factors for antibiotic resistance by
virtue of previous contact with the healthcare system and/or prior antibiotic exposure (see
Recommendation 10) and are therefore recommended to receive broader-spectrum therapy to
ensure adequate coverage. In addition to H. influenzae and M. catarrhalis (both of which
frequently produce b-lactamase), S. aureus and gram-negative bacilli are more common causes
of CAP in patients with comorbidities, such as COPD.
Regimens recommended for patients with comorbidities include a b-lactam or cephalosporin in
combination with either a macrolide or doxycycline. These combinations should effectively
target macrolide- and doxycycline-resistant S. pneumoniae (as b-lactam resistance in S.
pneumoniae remains less common), in addition to b-lactamase–producing strains of H.
influenzae, many enteric gramnegative bacilli, most methicillinsusceptible S. aureus, and M.
pneumoniae and C. pneumoniae. The monotherapies listed also are effective against most
common bacterial pathogens.
Both sets of treatment recommendations contain multiple antibiotic options without specifying
a preference order. The choice between these options requires a risk–benefit assessment for
each individual patient, weighing local epidemiological data against specific risk factors that
increase the risk of individual choices, such as documented b-lactam or macrolide allergy,
cardiac arrhythmia (macrolides), vascular disease (fluoroquinolones), and history of infection
with Clostridium difficile. In particular, despite the concern regarding adverse events associated
with fluoroquinolones, the panel believed that fluoroquinolone therapy was justified for adults
with comorbidities and CAP managed in the outpatient setting. Reasons included the
performance of fluoroquinolones in numerous studies of outpatient CAP (70, 72, 75, 77, 80, 83)
and inpatient CAP (see inpatient CAP section), the very low resistance rates in common
bacterial causes of CAP, their coverage of both typical and atypical organisms, their oral
bioavailability, the convenience of monotherapy, and the relative rarity of serious adverse
events related to their use. However, there have been increasing reports of adverse events
related to fluoroquinolone use as summarized on the U.S. Food and Drug Administration
website (96).
Of note, we adopt the convention of prior guidelines to recommend that patients with recent
exposure to one class of antibiotics recommended above receive treatment with antibiotics
from a different class, given increased risk for bacterial resistance to the initial treatment
regimen. We also highlight that although patients with significant risk factors for CAP due to
MRSA or P. aeruginosa (see Recommendation 11) are uncommonly managed in the outpatient
setting, these patients may require antibiotics that include coverage for these pathogens.
Research needed in this area. There is a need for head-to-head prospective RCTs of outpatient
CAP treatment, comparing clinical outcomes, including treatment failure, need for subsequent
visits, hospitalization, time to return to usual activities and adverse events. Furthermore, the
prevalence of specific pathogens and their antimicrobial susceptibility patterns in outpatients
with pneumonia should be monitored. Newer agents, including lefamulin and omadacycline,
need further validation in the outpatient setting.
Question 9: In the Inpatient Setting, Which Antibiotic Regimens Are Recommended for Empiric
Treatment of CAP in Adults without Risk Factors for MRSA and P. aeruginosa?
Recommendation 9.1. In inpatient adults with nonsevere CAP without risk factors for MRSA or
P. aeruginosa (see Recommendation 11), we recommend the following empiric treatment
regimens (in no order of preference) (Table 4):
 combination therapy with a b-lactam (ampicillin1sulbactam 1.5–3 g every 6 h,
cefotaxime 1–2 g every 8 h, ceftriaxone 1–2 g daily, or ceftaroline 600 mg every 12 h)
and a macrolide (azithromycin 500 mg daily or clarithromycin 500 mg twice daily)
(strong recommendation, high quality of evidence), or
 monotherapy with a respiratory fluoroquinolone (levofloxacin 750 mg daily,
moxifloxacin 400 mg daily) (strong recommendation, high quality of evidence).
A third option for adults with CAP who have contraindications to both macrolides and
fluoroquinolones is:
 combination therapy with a b-lactam (ampicillin1sulbactam, cefotaxime, ceftaroline, or
ceftriaxone, doses as above) and doxycycline 100 mg twice daily (conditional
recommendation, low quality of evidence).
Summary of the evidence. Most randomized controlled studies of hospitalized adults with CAP
comparing b-lactam/macrolide therapy versus fluoroquinolone monotherapy were designed as
noninferiority trials and had limited sample sizes (97–103). These data suggested that patients
treated with b-lactam/macrolide therapy have similar clinical outcomes compared with those
treated with fluoroquinolone monotherapy. A systematic review of 16 RCTs in 4,809 patients
found fluoroquinolone monotherapy resulted in significantly fewer incidences of clinical failure,
treatment discontinuation, and diarrhea than b-lactam/macrolide combination (104). However,
mortality rates were low overall, and there were no significant differences in mortality between
groups. Another systematic review of 20 experimental and observational studies in adults
hospitalized with radiographically confirmed CAP, b-lactam plus macrolide combination
therapy, or fluoroquinolone monotherapy were generally associated with lower mortality than
b-lactam monotherapy (105). Therefore, the panel recommends a b-lactam (ampicillin plus
sulbactam, cefotaxime, ceftaroline, or ceftriaxone) plus macrolide (azithromycin or
clarithromycin) or monotherapy with a respiratory fluoroquinolone (levofloxacin, moxifloxacin)
for the management of inpatients with nonsevere CAP. (Of note, azithromycin but not
clarithromycin is available in parenteral formulation.) In choosing between these two options,
clinicians should weigh the risks and benefits of the drugs, particularly in light of individual risk
factors, such as a history of C. difficile infection or risk factors related to U.S. Food and Drug
Administration warnings (96). The panel recommends using doxycycline as an alternative to a
macrolide in combination with a blactam as a third option in the presence of documented
allergies or contraindications to macrolides or fluoroquinolones or clinical failure on one of
those agents. Of note, a newer member of the tetracycline class, omadacycline, was recently
reported to be equivalent to moxifloxacin as monotherapy for adults with nonsevere CAP and is
effective in the setting of tetracycline resistance (106). However, as this is a single published
report and the safety information is less well established, the committee decided to not list this
new agent as an alternative to the currently recommended treatment options.
The panel also considered b-lactam monotherapy as an option for inpatients with nonsevere
CAP. An RCT in 580 patients with CAP could not rule out the possibility that b-lactam
monotherapy was inferior to b-lactam/macrolide therapy for inpatients with CAP (107). Nie and
colleagues identified several cohort (n=4) and retrospective observational (n = 12) studies
addressing this question and found that b-lactam/macrolide therapy reduced mortality in
patients with CAP compared with patients treated with b-lactam monotherapy (108). Similarly,
Horita and colleagues demonstrated that b-lactam/macrolide combinations may decrease all-
cause death, but mainly for patients with severe CAP (109). Therefore, we suggest that b-
lactam monotherapy should not be routinely used for inpatients with CAP over fluoroquinolone
monotherapy or b-lactam/macrolide combination therapy.
Rationale for the recommendation. As summarized in Table 4, the empiric antibiotic coverage
recommendations for patients hospitalized with CAP remain aligned to cover the most likely
pathogens causing CAP. There is a paucity of RCTs to favor the recommendation of combination
b-lactam plus macrolide versus monotherapy with a respiratory fluoroquinolone versus
combined therapy with b-lactam plus doxycycline.
Research needed in this area. There is a need for higher-quality evidence in support of the use
of combination therapy with a b-lactam and doxycycline. Given concerns over increasing drug
resistance (macrolides) and safety issues (macrolides, fluoroquinolones), there is a need for
research on new therapeutic agents for adults with CAP including omadacycline (see above)
and lefamulin, a new pleuromutilin antibiotic that was recently demonstrated to be noninferior
to moxifloxacin in hospitalized adult patients with CAP (110).
Recommendation 9.2. In inpatient adults with severe CAP (see Table 1) without risk factors for
MRSA or P. aeruginosa, we recommend (Table 4) (note, specific agents and doses are the same
as 9.1):
 a b-lactam plus a macrolide (strong recommendation, moderate quality of evidence); or
 a b-lactam plus a respiratory fluoroquinolone (strong recommendation, low quality of
evidence).
Summary of the evidence. In the absence of RCTs evaluating therapeutic alternatives in severe
CAP, the evidence is from observational studies that used different definitions of illness severity
to address this question. Sligl and colleagues found in a meta-analysis of observational studies
with almost 10,000 critically ill patients with CAP that macrolidecontaining therapies (often in
combination with a b-lactam) were associated with a significant mortality reduction (18%
relative risk, 3% absolute risk) compared with non–macrolide-containing therapies (111). A
mortality benefit from macrolides has been observed mainly in cohorts with a large number of
patients with
severe CAP. In a systematic review, Vardakas and colleagues compared a
b-lactam/fluoroquinolone versus a b-lactam/macrolide combination for the treatment of
patients with CAP (112). The authors found 17 observational studies and no RCTs addressing
this comparison. The combination of b-lactam/fluoroquinolone therapy was associated with
higher mortality than b-lactam/macrolide combination therapy, but the overall quality of the
studies was judged to be low, precluding a definitive recommendation (112).
Rationale for the recommendation. In the absence of data from clinical trials demonstrating the
superiority of any specific regimen for patients with severe CAP, the committee considered
epidemiological data for severe CAP pathogens and observational studies comparing different
regimens. As a result, we recommend that combination therapy with a b-lactam plus a
macrolide or a blactam plus a respiratory fluoroquinolone should be the treatment of choice for
patients with severe CAP. Both fluoroquinolone monotherapy and the combination of b-lactam
plus doxycycline have not been well studied in severe CAP and are not recommended as
empiric therapy for adults with severe CAP.
Research needed in this area. Future well-designed RCTs should focus on therapies for patients
at highest risk of death with severe pneumonia, as these are needed to assess the benefits and
risks of combination b-lactam and macrolide therapy compared with b-lactam and respiratory
fluoroquinolone therapy. Studies of fluoroquinolone monotherapy in severe CAP are also
needed.

Question 10: In the Inpatient Setting, Should Patients with Suspected Aspiration Pneumonia
Receive Additional Anaerobic Coverage beyond Standard Empiric Treatment for CAP?
Recommendation. We suggest not routinely adding anaerobic coverage for suspected
aspiration pneumonia unless lung abscess or empyema is suspected (conditional
recommendation, very low quality of evidence).
Summary of the evidence. Aspiration is a common event, and as many as half of all adults
aspirate during sleep (113). As a result, the true rate of aspiration pneumonia is difficult to
quantify, and there is no definition that separates patients with aspiration pneumonia from all
others diagnosed with pneumonia. Some have estimated that 5% to 15% of pneumonia
hospitalizations are associated with aspiration (114). Rates are higher in populations admitted
from nursing homes or extended care facilities (115). Patients who aspirate gastric contents are
considered to have aspiration pneumonitis. Many of these patients have resolution of
symptoms within 24 to 48 hours and require only supportive treatment, without antibiotics
(116).
Studies evaluating the microbiology of patients with aspiration pneumonia in the 1970s showed
high rates of isolation of anaerobic organisms (117, 118); however, these studies often used
trans-tracheal sampling and evaluated patients late in their disease course, two factors that
may have contributed to a higher likelihood of identifying anaerobic organisms (114). Several
studies of acute aspiration events in hospitalized patients have suggested that anaerobic
bacteria do not play a major role in etiology (119–121).
With increasing rates of C. difficile infections (frequently associated with use of clindamycin),
the question of adding empiric anaerobic coverage (clindamycin or b-lactam/b-lactamase
inhibitors) in addition to routine CAP treatment in patients with suspected aspiration is an
important one. However, there are no clinical trials comparing treatment regimens with and
without anaerobic coverage for patients hospitalized with suspected aspiration. Most recent
studies are small, retrospective, and provide only observational data on microbiologic patterns
and treatment regimens for patients hospitalized with suspected aspiration pneumonia.
Rationale for the recommendation. Although older studies of patients with aspiration
pneumonia showed high isolation rates of anaerobic organisms, more recent studies have
shown that anaerobes are uncommon in patients hospitalized with suspected aspiration (119,
120). Increasing prevalence of antibiotic-resistant pathogens and complications of antibiotic use
highlight the need for a treatment approach that avoids unnecessary use of antibiotics.
Research needed in this area. Clinical trials evaluating diagnostic and treatment strategies in
patients with suspected aspiration are needed, especially in terms of the ability to distinguish
micro- and macroaspiration events that lead to lower respiratory tract infection from those that
do not result in infection.

Question 11: In the Inpatient Setting, Should Adults with CAP and Risk Factors for MRSA or P.
aeruginosa Be Treated with Extended-Spectrum Antibiotic Therapy Instead of Standard CAP
Regimens?
Recommendation. We recommend abandoning use of the prior categorization of healthcare-
associated pneumonia (HCAP) to guide selection of extended antibiotic coverage in adults with
CAP (strong recommendation, moderate quality of evidence).
We recommend clinicians only cover empirically for MRSA or P. aeruginosa in adults with CAP if
locally validated risk factors for either pathogen are present (strong recommendation,
moderate quality of evidence). Empiric treatment options for MRSA include vancomycin (15
mg/kg every 12 h, adjust based on levels) or linezolid (600 mg every 12 h). Empiric treatment
options for P. aeruginosa include piperacillin-tazobactam (4.5 g every 6 h), cefepime (2 g every
8 h), ceftazidime (2 g every 8 h), aztreonam (2 g every 8 h), meropenem (1 g every 8 h), or
imipenem (500 mg every 6 h).
If clinicians are currently covering empirically for MRSA or P. aeruginosa in adults with CAP on
the basis of published risk factors but do not have local etiological data, we recommend
continuing empiric coverage while obtaining culture data to establish if these pathogens are
present to justify continued treatment for these pathogens after the first few days of empiric
treatment (strong recommendation, low quality of evidence).
Summary of the evidence. HCAP, as a distinct clinical entity warranting unique antibiotic
treatment, was incorporated into the 2005 ATS/IDSA guidelines for management of hospital-
acquired and ventilator-associated pneumonia (122). HCAP was defined for those patients who
had any one of several potential risk factors for antibiotic-resistant pathogens, including
residence in a nursing home and other long-term care facilities, hospitalization for >2 days in
the last 90 days, receipt of home infusion therapy, chronic dialysis, home wound care, or a
family member with a known antibiotic-resistant pathogen. The introduction of HCAP was
based on studies identifying a higher prevalence of pathogens that are not susceptible to
standard first-line antibiotic therapy, in particular MRSA and P. aeruginosa, in some subsets of
patients with CAP (123). Since then, many studies have demonstrated that the factors used to
define HCAP do not predict high prevalence of antibiotic-resistant pathogens in most settings.
Moreover, a significant increased use of broad-spectrum antibiotics (especially vancomycin and
antipseudomonal b-lactams) has resulted, without any apparent improvement in patient
outcomes (124–133).
Studies have identified risk factors for antibiotic-resistant pathogens, and in some cases the risk
factors are distinct for MRSA versus P. aeruginosa (134–154). However, most of these individual
risk factors are weakly associated with these pathogens. The most consistently strong individual
risk factors for respiratory infection with MRSA or P. aeruginosa are prior isolation of these
organisms, especially from the respiratory tract, and/or recent hospitalization andexposure to
parenteral antibiotics (134, 155, 156). Therefore, we have highlighted these individual risk
factors to help guide initial microbiological testing and empiric coverage for these pathogens.
Unfortunately, no validated scoring systems exist to identify patients with MRSA or P.
aeruginosa with sufficiently high positive predictive value to determine the need for empiric
extended-spectrum antibiotic treatment. Scoring system development and validation are
complicated by varying prevalence of MRSA and P. aeruginosa in different study populations.
Moreover, no scoring system has been demonstrated to improve patient outcomes or reduce
overuse of broad-spectrum antibiotics.
Although there is limited evidence to support the use of a specific set of risk factors to identify
patients with sufficiently high risk of MRSA or P. aeruginosa to warrant extended-spectrum
therapy, a stronger evidence base guides deescalation of therapy after extendedspectrum
therapy is initially prescribed. Although no randomized prospective studies have been reported,
recent observational (157) and retrospective (158–161) studies in patients with CAP provide
strong evidence that deescalation of antibiotic therapy at 48 hours in accord with
microbiological results that do not yield MRSA or P. aeruginosa is safe and reduces duration of
antibiotic treatment, length of hospitalization, and complications of broad-spectrum therapy.
These results are reinforced by retrospective (162) and prospective and randomized but not
blinded (163) studies of patients with severe sepsis, the majority of whom had pneumonia, as
well as by a recent meta-analysis of deescalation in adults with sepsis (164).
We propose that clinicians need to obtain local data on whether MRSA or P. aeruginosa is
prevalent in patients with CAP and what the risk factors for infection are at a local (i.e., hospital
or catchment area) level. We refer to this process as “local validation.” This recommendation is
based on the absence of high-quality outcome studies, the very low prevalence of MRSA or P.
aeruginosa in most centers, and significant increased use of anti-MRSA and antipseudomonal
antibiotics for treatment of CAP (142, 155, 165). Although we acknowledge that centers may
not currently have local prevalence data, adopting the recommendations for culture of sputum
and blood when risk factors for MRSA or P. aeruginosa are present will enable clinicians to
generate these local data over time. We recommend analyzing the frequency of MRSA or P.
aeruginosa as a CAP pathogen relative to the number of all cases of CAP, not just those for
whom cultures are sent. Finally, routine cultures in patients empirically treated for MRSA or P.
aeruginosa allow deescalation to standard CAP therapy if cultures do not reveal a drug-resistant
pathogen and the patient is clinically improving at 48 hours.
Rationale for the recommendation. Our approach to treating inpatient adults with CAP is
summarized in Table 4. Our recommendation against using the former category of HCAP as a
basis for selecting extended-spectrum therapy is based on high-quality studies of patient
outcomes. Although we understand that clinicians would prefer a simple rule that does not
require incorporating site-specific data, the current evidence does not permit endorsement of a
simple and accurate rule to determine which patients with CAP should be covered for MRSA
and/or P. aeruginosa. However, the alternative approach to MRSA and P. aeruginosa that we
propose as a replacement is not based on high-quality studies, because such studies do not
exist. The lack of adequate outcome data and marked variation between sites in prevalence of
MRSA and P. aeruginosa make generalizing any findings extremely difficult. We hope that
future research will improve our understanding of this challenging clinical problem.
Our first principle was to maintain the distinction between severe and nonsevere pneumonia as
per prior guidelines, because the risk of inadequate empiric antibiotic therapy is much greater
in severe CAP. As noted previously, severity is defined by the degree of physiological
impairment, as classified by the IDSA/ATS 2007 criteria.
The second principle was that there is sufficient evidence that prior identification of MRSA or P.
aeruginosa in the respiratory tract within the prior year predicts a very high risk of these
pathogens being identified in patients presenting with CAP (139, 141, 143, 150, 155, 165), and
therefore these were sufficient indications to recommend blood and sputum cultures and
empiric therapy for these pathogens in patients with CAP in addition to coverage for standard
CAP pathogens, with deescalation at 48 hours if cultures are negative. We endorse the empiric
treatment recommendations for MRSA and P. aeruginosa provided by the 2016 Clinical Practice
Guideline from IDSA and ATS for the management of adults with hospitalacquired and
ventilator-associated pneumonia (166).
The major additional risk factors for MRSA and P. aeruginosa identified in the literature are
hospitalization and parenteral antibiotic exposure in the last 90 days (136–138, 140, 142–151,
153). In patients with recent hospitalization and exposure to parenteral antibiotics, we
recommend microbiological testing without empiric extended-spectrum therapy for treatment
of nonsevere CAP and microbiological testing with extended-spectrum empiric therapy in
addition to coverage for standard CAP pathogens for treatment of severe CAP, with
deescalation at 48 hours if cultures are negative and the patient is improving.
The data supporting rapid MRSA nasal testing are robust (167, 168), and treatment for MRSA
pneumonia can generally be withheld when the nasal swab is negative, especially in nonsevere
CAP. However, the positive predictive value is not as high; therefore, when the nasal swab is
positive, coverage for MRSA pneumonia should generally be initiated, but blood and sputum
cultures should be sent and therapy deescalated if cultures are negative. However, this latter
strategy of de-escalation. in the face of a positive nasal swab will vary depending on the
severity of CAP and the local prevalence of MRSA as a pathogen.

Question 12: In the Inpatient Setting, Should Adults with CAP Be Treated with Corticosteroids?
Recommendation. We recommend not routinely using corticosteroids in adults with nonsevere
CAP (strong recommendation, high quality of evidence).
We suggest not routinely using corticosteroids in adults with severe CAP (conditional
recommendation, moderate quality of evidence).
We suggest not routinely using corticosteroids in adults with severe influenza pneumonia
(conditional recommendation, low quality of evidence).
We endorse the Surviving Sepsis Campaign recommendations on the use of corticosteroids in
patients with CAP and refractory septic shock (169).
Summary of the evidence. Two randomized controlled studies of corticosteroids used for
treatment of CAP have shown significant reductions in mortality, length of stay, and/or organ
failure. The first study found a large magnitude of mortality benefit that has not been replicated
in other studies, raising concerns that the results overestimated the true effect (170). In the
second study, there were baseline differences in renal function between groups (171). Other
RCTs of corticosteroids in the treatment of CAP have not shown significant differences in
clinically important endpoints. Differences have been observed in the time to resolution of
fever and other features of clinical stability, but these have not translated into differences in
mortality, length of stay, or organ failure (172, 173).
Some (174, 175), but not all (176, 177), meta-analyses of published corticosteroid studies have
shown a mortality benefit in patients with severe CAP, although no consistent definition of
disease severity was used. Side effects of corticosteroids (on the order of 240 mg of
hydrocortisone per day) include significant increases in hyperglycemia requiring therapy and
possible higher secondary infection rates (178, 179). No reported study has shown excess
mortality in the corticosteroidtreated group.
In pneumonia due to influenza, a metaanalysis (180) of predominantly small retrospective
studies suggests that mortality may be increased in patients who receive corticosteroids. This
finding might reflect the importance of innate immunity in defense against influenza as
opposed to bacterial pneumonia.
Rationale for the recommendation. There are no data suggesting benefit of corticosteroids in
patients with nonsevere CAP with respect to mortality or organ failure and only limited data in
patients with severe CAP. The risk of corticosteroids in the dose range up to 240 mg of
hydrocortisone equivalent per day for a maximum of 7 days is predominantly hyperglycemia,
although rehospitalization rates may also be higher (176), and more general concerns about
greater complications in the following 30 to 90 days have been raised (179). At least one large
trial (clinicaltrials.gov NCT01283009) has been completed but not reported and may further
inform which subgroups of patients benefit from steroids. We also endorse the Surviving Sepsis
Campaign recommendations on the use of steroids in patients with septic shock refractory to
adequate fluid resuscitation and vasopressor support (169).
Of note, there is no intent that our recommendations would override clinically appropriate use
of steroids for comorbid diseases, such as chronic obstructive pulmonary disease, asthma, and
autoimmune diseases, where corticosteroids are supported as a component of treatment.
Research needed in this area. Large, multicenter, randomized trials with welldefined inclusion
and exclusion criteria and measurement of multiple relevant clinical outcomes are needed to
define the subsets of patients (if any) who benefit or are potentially harmed from corticosteroid
therapy. The trial should also make extensive efforts to define causative pathogens, to define
whether there are clear pathogen-specific indications or contraindications for corticosteroid
therapy (especially disease due to S. pneumoniae and influenza).

Question 13: In Adults with CAP Who Test Positive for Influenza, Should the Treatment Regimen
Include Antiviral Therapy?
Recommendation. We recommend that antiinfluenza treatment, such as oseltamivir, be
prescribed for adults with CAP who test positive for influenza in the inpatient setting,
independent of duration of illness before diagnosis (strong recommendation, moderate quality
of evidence).
We suggest that antiinfluenza treatment be prescribed for adults with CAP who test positive for
influenza in the outpatient setting, independent of duration of illness before diagnosis
(conditional recommendation, low quality of evidence).
Summary of the evidence. No clinical trials have evaluated the effect of treatment with
antiinfluenza agents in adults with influenza pneumonia, and data are lacking on the benefits of
using antiinfluenza agents in the outpatient setting for patients with CAP who test positive for
influenza virus. Several observational studies suggest that treatment with oseltamivir is
associated with reduced risk of death in patients hospitalized for CAP who test positive for
influenza virus (181, 182). Treatment within 2 days of symptom onset or hospitalization may
result in the best outcomes (183, 184), although there may be benefits up to 4 or 5 days after
symptoms begin (181, 185).
The use of antiinfluenza agents in the outpatient setting reduces duration of symptoms and the
likelihood of lower respiratory tract complications among patients with influenza (186), with
most benefit if therapy is received within 48 hours after onset of symptoms (187).
Rationale for the recommendation. For inpatients, a substantial body of observational evidence
suggests that giving antiinfluenza agents reduces mortality risk in adults with influenza
infection. Although benefits are strongest when therapy is started within 48 hours of symptom
onset, studies also support starting later (185). These data underlie our strong recommendation
for using antiinfluenza agents for patients with CAP and influenza in the inpatient setting,
consistent with the recently published IDSA Influenza Clinical Practice Guideline (33).
Although we did not identify studies that specifically evaluated antiinfluenza agents for treating
outpatients with CAP who test positive for influenza, we make the same recommendation as for
inpatients, on the basis of the inpatient data and on outpatient data showing better time to
resolution of symptoms and prevention of hospitalization among those with influenza but
without pneumonia. Our recommendations are consistent with the NTIDSA influenza
guidelines, which were recently released (33).
Research needed in this area. Randomized controlled studies are needed to support the
recommendation for use of antiinfluenza agents to treat for influenza pneumonia in the
outpatient setting. In particular, knowing whether therapy is valuable when started more the 48
hours after symptom onset would help guide clinical decision-making.

Question 14: In Adults with CAP Who Test Positive for Influenza, Should the Treatment Regimen
Include Antibacterial Therapy?
Recommendation. We recommend that standard antibacterial treatment be initially prescribed
for adults with clinical and radiographic evidence of CAP who test positive for influenza in the
inpatient and outpatient settings (strong recommendation, low quality of evidence).
Summary of the evidence. Bacterial pneumonia can occur concurrently with influenza virus
infection or present later as a worsening of symptoms in patients who were recovering from
their primary influenza virus infection. As many as 10% of patients hospitalized for influenza
and bacterial pneumonia die as a result of their infection (188). An autopsy series from the
2009 H1N1 influenza pandemic found evidence of bacterial coinfection in about 30% of deaths
(189).
S. aureus is one of the most common bacterial infections associated with influenza pneumonia,
followed by S. pneumoniae, H. influenzae, and group A Streptococcus; other bacteria have also
been implicated (188, 190–192). Given this spectrum of pathogens, appropriate agents for
initial therapy include the same agents generally recommended for CAP. Risk factors and need
for empiric coverage for MRSA would follow the guidelines included earlier in this document.
Rapidly progressive severe pneumonia with MRSA has been described in previously healthy
young patients, particularly in the setting of prior influenza; however, it is typically readily
identified in the nares or sputum and should be identified by following the recommendations of
earlier recommendations in this guideline.
Rationale for the recommendation. The recommendation to routinely prescribe antibacterial
agents in patients with influenza virus infection and pneumonia was based on evidence
suggesting that bacterial coinfections are a common and serious complication of influenza, as
well as the inability to exclude the presence of bacterial coinfection in a patient with CAP who
has a positive test for influenza virus. Although low levels of biomarkers such as procalcitonin
decrease the likelihood that patients have bacterial infections, these biomarkers do not
completely rule out bacterial pneumonia in an individual patient with sufficient accuracy to
justify initially withholding antibiotic therapy, especially among patients with severe CAP (37,
38, 193). We have provided a strong recommendation because of the significant risk of
treatment failure in delaying appropriate antibacterial therapy in patients with CAP. However in
patients with CAP, a positive influenza test, no evidence of a bacterial pathogen (including a low
procalcitonin level), and early clinical stability, consideration could be given to earlier
discontinuation of antibiotic treatment at 48 to 72 hours.
Research needed in this area. Randomized controlled studies are needed to establish whether
antibacterial therapy can be stopped at 48 hours for patients with CAP who test positive for
influenza and have no biomarker (e.g., procalcitonin) or microbiological evidence of a
concurrent bacterial infection.

Question 15: In Outpatient and Inpatient Adults with CAP Who Are Improving, What Is the
Appropriate Duration of Antibiotic Treatment?
Recommendation. We recommend that the duration of antibiotic therapy should be guided by
a validated measure of clinical stability (resolution of vital sign abnormalities [heart rate,
respiratory rate, blood pressure, oxygen saturation, and temperature], ability to eat, and
normal mentation), and antibiotic therapy should be continued until the patient achieves
stability and for no less than a total of 5 days (strong recommendation, moderate quality of
evidence).
Summary of the evidence. A small number of randomized trials address the appropriate
duration of antibiotic therapy in CAP, and randomized placebo-controlled trials of high quality
are mostly limited to the inpatient setting. In these trials, no difference was observed between
5 additional days of oral amoxicillin compared with placebo in patients who had clinically
improved on 3 days of intravenous amoxicillin (194), or between 2 days of intravenous
cefuroxime followed by 5 days versus 8 days of oral cefuroxime (195). Similar results were
obtained with 5 days of levofloxacin 750 mg daily compared with 10 days of levofloxacin 500
mg daily (196) and 5 days of intravenous ceftriaxone compared with 10 days (197). Several
recent metaanalyses similarly demonstrate the efficacy of shorter courses of antibiotic therapy
of 5 to 7 days (198–200).
Several studies have demonstrated that the duration of antibiotic therapy can be reduced in
patients with CAP with the use of a procalcitonin-guided pathway and serial procalcitonin
measurement compared with conventional care, but in most cases the average length of
treatment was greatly in excess of current U.S. standards of practice as well as the
recommendations of these current guidelines. Concern has also been raised that procalcitonin
levels may not be elevated when there is concurrent viral and bacterial infection (201, 202) or
with important pathogens such as Legionella and Mycoplasma spp (37, 201, 203). Serial
procalcitonin measurement is therefore likely to be useful primarily in settings where the
average length of stay for patients with CAP exceeds normal practice (e.g., 5–7 d).
It is recognized that some patients do not respond to a standard duration of therapy. A variety
of criteria for determining clinical improvement have been developed for patients with CAP and
validated in clinical trials, including resolution of vital sign abnormalities (heart rate, respiratory
rate, blood pressure, oxygen saturation, and temperature), ability to eat, and normal mentation
(204). Failure to achieve clinical stability within 5 days is associated with higher mortality and
worse clinical outcomes (205–207). Such failure should prompt assessment for a pathogen
resistant to the current therapy and/or complications of pneumonia (e.g., empyema or lung
abscess) or for an alternative source of infection and/or inflammatory response (208, 209).
When assessment of clinical stability has been introduced into clinical practice, patients have
shorter durations of antibiotic therapy without adverse impact on outcome (210). All clinicians
should therefore use an assessment of clinical stability as part of routine care of patients with
CAP.
Longer courses of antibiotic therapy are recommended for 1) pneumonia Scomplicated by
meningitis, endocarditis, and other deep-seated infection; or 2) infection with other, less-
common pathogens not covered in these guidelines (e.g., Burkholderia pseudomallei,
Mycobacterium tuberculosis or endemic fungi).
Rationale for the recommendation. As recent data supporting antibiotic administration for ,5
days are scant, on a risk–benefit basis we recommend treating for a minimum of 5 days, even if
the patient has reached clinical stability before 5 days. As most patients will achieve clinical
stability within the first 48 to 72 hours, a total duration of therapy of 5 days will be appropriate
for most patients. In switching from parenteral to oral antibiotics, either the same agent or the
same drug class should be used.
We acknowledge that most studies in support of 5 days of antibiotic therapy include patients
without severe CAP, but we believe these results apply to patients with severe CAP and without
infectious complications. We believe that the duration of therapy for CAP due to suspected or
proven MRSA or P. aeruginosa should be 7 days, in agreement with the recent hospitalacquired
pneumonia and ventilatorassociated pneumonia guidelines (166).
Research needed in this area. Controlled studies are needed to establish the duration of
antibiotic therapy for adults with complications of CAP, including empyema, and adults with
prolonged time to achieving clinical stability.

Question 16: In Adults with CAP Who Are Improving, Should Follow-up Chest Imaging Be
Obtained?
Recommendation. In adults with CAP whose symptoms have resolved within 5 to 7 days, we
suggest not routinely obtaining follow-up chest imaging (conditional recommendation, low
quality of evidence).
Summary of the evidence. There are limited data on the clinical usefulness of reimaging
patients with pneumonia. Most available data have evaluated whether reimaging patients
detects lung malignancy not recognized at the time of treatment for pneumonia. Reported
rates of malignancy in patients recovering from CAP range from 1.3% to 4% (211–214). When
unsuspected nonmaligant pathology is included, the rate of abnormal findings may reach 5%.
Almost all patients with malignancy in reported series were smokers or ex-smokers. One
longer-term study found 9.2% of CAP survivors in the Veterans Affairs system (with a
predominantly male population and high smoking prevalence) had a new diagnosis of cancer,
with a mean time to diagnosis of 297 days. However, only 27% were diagnosed within 90 days
of discharge from hospital, suggesting the yield of routine follow-up post discharge would be
low (215).
Rationale for the recommendation. Available data suggest the positive yield from repeat
imaging ranges from 0.2% to 5.0%; however many patients with new abnormalities in these
studies meet criteria for lung cancer screening among current or past smokers (216).
Research needed in this area. Further research may clarify subgroups of patients who may
benefit from further radiological assessment after initial therapy for pneumonia.

Conclusions
Recommendations to help clinicians optimize therapy for their patients with CAP have been
revised in light of new data.
Methods of quality improvement are critical to the implementation of guideline
recommendations. It remains disappointing how few key clinical questions have been studied
adequately enough to allow for strong recommendations regarding the standard of care. We
hope that the research priorities outlined in this document will prompt new investigations
addressing key knowledge gaps.
Despite substantial concern over the rise of antibiotic-resistant pathogens, most patients with
CAP can be adequately treated with regimens that have been used for multiple decades. It is
also true that the subset of patients with CAP who have significant comorbidities and frequent
contact with healthcare settings and antibiotics is increasing, and, in some settings, the rates of
infection with MRSA or P. aeruginosa are high enough to warrant empiric treatment.
Unfortunately, microbiological testing has yet to deliver fast, accurate, and affordable testing
that results in proven benefit for patients with CAP in terms of more rapid delivery of targeted
therapy or safe deescalation of unnecessary therapy. Exceptions include rapid testing for MRSA
and influenza. Until we have such widely available (and affordable) tests, therapy for many or
most patients with CAP will remain empiric. Therefore, clinicians need to be aware of the
spectrum of local pathogens, especially if they care for patients at a center where infection with
antibiotic-resistant pathogens such as MRSA and P. aeruginosa are more common.
A difference between this guideline and previous ones is that we have significantly increased
the proportion of patients in whom we recommend routinely obtaining respiratory tract
samples for microbiologic studies. This decision is largely based on a desire to correct the
overuse of anti-MRSA and antipseudomonal therapy that has occurred since the introduction of
the HCAP classification (which we recommend abandoning) rather than high-quality evidence.
We expect this change will generate significant research to prove or disprove the value of this
approach. As it is not possible to create a “one size fits all” schema for empiric therapy for CAP,
clinicians must validate any approach taking into account their local spectrum and frequency of
resistant pathogens, which is another driver for recommending increased testing. We similarly
expect our move against endorsing monotherapy with macrolides, which is based on
population resistance data rather than high-quality clinical studies, will generate future
outcomes studies comparing different treatment strategies.
We hope that clinicians and researchers will find this guideline useful, but the
recommendations included here do not obviate the need for clinical assessment and knowledge
to ensure each individual patient receives appropriate and timely care. However, this guideline
delineates minimum clinical standards that are achievable and will help drive the best patient
outcomes on the basis of currently available data.