Community-acquired pneumonia in adults: Assessing severity and determining the appropriate site of care

INTRODUCTION — Community-acquired pneumonia (CAP) is a common and potentially serious illness and one of the most common causes of sepsis. It is associated with considerable morbidity and mortality, particularly in older adult patients and those with coexisting comorbidities. (See "Morbidity and mortality associated with community-acquired pneumonia in adults".)

Hospital admission rates in the United States for adults with CAP vary widely and are often not directly related to local disease severity. This suggests that clinicians use inconsistent criteria when making the initial decision about the appropriate site of care. In addition, clinicians often overestimate patient risk of short-term mortality, even among low-risk patients [1]. A likely consequence of these overestimates is unnecessary admissions; low-risk patients with a physician-estimated risk of death in excess of 5 percent were over six times more likely to be hospitalized after adjusting for other potential confounders of hospitalization [1].

This topic review focuses on clinical prediction rules and practice guidelines to quantify illness severity and determine the initial site of treatment for CAP, thereby avoiding unnecessary hospitalizations.

Other important issues related to CAP are discussed separately, including pneumonia as part of coronavirus disease 2019 (COVID-19):

●(See "Clinical evaluation and diagnostic testing for community-acquired pneumonia in adults".)

●(See "Treatment of community-acquired pneumonia in adults who require hospitalization".)

●(See "Treatment of community-acquired pneumonia in adults in the outpatient setting".)

●(See "Epidemiology, pathogenesis, and microbiology of community-acquired pneumonia in adults".)

●(See "Morbidity and mortality associated with community-acquired pneumonia in adults".)

●(See "Aspiration pneumonia in adults" and "Epidemiology of pulmonary infections in immunocompromised patients".)

●(See "COVID-19: Evaluation of adults with acute illness in the outpatient setting" and "COVID-19: Management in hospitalized adults" and "COVID-19: Management in nursing homes" and "COVID-19: Management of adults with acute illness in the outpatient setting".)

APPROACH TO SITE OF CARE — Severity of illness is the most critical factor in making the site of care decision once pneumonia is identified, along with other factors. These additional influences on site of care include the ability to maintain oral intake, likelihood of medication adherence, history of active substance use disorder, mental illness, cognitive or functional impairment, and living or social circumstances (eg, homelessness, residence far enough from a health care facility that precludes timely return to care in the event of clinical worsening). Although a definitive etiologic pneumonia diagnosis is often not established until after the site of treatment decision is needed, severity of illness and other factors at presentation plus clinical or epidemiologic evidence favoring pathogens associated with rapidly progressive forms of pneumonia (eg, postinfluenza bacterial pneumonias, severe acute respiratory syndrome, Middle East respiratory syndrome, avian influenza [eg, H5N1, H7N9], Legionella pneumonia, COVID-19) will inform the need for close clinical follow-up particularly for patients initially deemed low risk at presentation and treated outside of the hospital setting.

Our approach for determining the need for admission starts with objective assessment of the severity of illness and is summarized below and in the following algorithm (algorithm 1):

●If the patient has septic shock or respiratory failure, transfer promptly to a hospital critical care site (ICU or emergency department). (See 'Admission to intensive care' below.)

●If there are practical concerns about outpatient management (such as those outlined above) or the patient's oxygen saturation is <92 percent on room air and represents a significant change from baseline, hospital admission is warranted. (See 'Importance of oxygen saturation' below.)

●If the two features above are not present, use a validated clinical prediction rule for prognosis to supplement, not replace, physician judgment.

•Among the available prediction rules, we prefer the Pneumonia Severity Index (PSI) because it is the most accurate and its safety and effectiveness in guiding clinical decision-making have been empirically confirmed (calculator 1). (See 'Pneumonia Severity Index' below.)

-Patients who are PSI class I or II can be managed as outpatients. Some PSI class II patients may benefit from in-home health care support, also termed "hospital-at-home" (eg, a visiting nurse, intravenous fluids, intravenous antibiotics). (See 'Additional options for site of care' below.)

-Patients who are PSI class III can usually be managed as outpatients, as outpatients with in-home health care support, or with a brief (<23 hour) observational stay. The choice of site of care will depend on each patient's clinical status, local resources, and other circumstances. (See 'Additional options for site of care' below.)

-Admit patients who are PSI class IV or V to the hospital. ICU admission may be needed for patients who require hemodynamic, ventilatory, or other organ support.

•The CURB-65 and other derivative scoring systems are alternatives that can be used when a less complex prediction rule for prognosis is desired (calculator 2). These rules have not had the safety and effectiveness assessments in guiding the initial site of treatment have like the PSI. (See 'The CURB-65 and CRB-65 scores' below.)

-Patients with a CURB-65 score of 0 can be managed as outpatients.

-Patients with a CURB-65 score of 1 or 2 should generally be admitted to a general medical ward, although patients who receive a score of 1 because they are ≥65 years of age who do not have major comorbidities do not necessarily require hospital admission. Such patients may also be managed as outpatients with in-home health care support, or with a brief (<23 hour) observational stay.

-Patients with a CURB-65 score of 3 to 5 should be admitted to the hospital and consider ICU admission based on the factors described above.

●For hospitalized patients, the clinician may seek additional markers of illness severity that support higher acuity levels of inpatient care. Patients with ≥3 of the Infectious Diseases Society of America/American Thoracic Society minor severity criteria often warrant admission to higher acuity levels of inpatient care, including the ICU. (See 'IDSA/ATS severity criteria' below.)

CLINICAL PREDICTION RULES FOR SEVERITY — Since 1987, numerous studies identified independent predictors of adverse medical outcomes for the purpose of objectively assessing the severity of illness for patients presenting with community-acquired pneumonia (CAP) [2-17]. We review the quality of these prediction rules for application in emergency department and ambulatory care settings using accepted evaluation standards (ie, relevant patient population, predictors, outcomes, study sites, and design; appropriate modelling techniques used; accuracy of rule assessed; and clinical effects tested) [18]. Recommendations related to the use of these prediction rules are provided above. (See 'Approach to site of care' above.)

Overall, 15 prediction rules satisfied five or fewer standards for quality [2,4,6-15]. One rule satisfied six standards, although the criterion for determination of accuracy was weakened by the use of just 71 patients [5]. The Pneumonia Severity Index (PSI) satisfied all seven quality standards for prediction rules and was the only one shown to be safe and effective in guiding clinical practice (eighth standard).

Pneumonia Severity Index — The PSI was derived and validated as part of the Pneumonia Patient Outcomes Research Team (PORT) prospective cohort study for the purpose of identifying patients with CAP at low risk for mortality [3].

The PSI stratified adults with radiographic evidence of pneumonia into five classes for risk of death from all causes within 30 days of presentation. Predictor variables were derived from the medical history, physical examination, and selected laboratory and radiographic findings readily available at the time of patient presentation. In contrast with previous prediction rules, the PSI application uses two steps that parallel decision-making processes the physician usually follows during a patient encounter. Step 1 of the rule identifies patients in the lowest risk class (risk class I) without using laboratory tests based upon age and the absence of five comorbid conditions and five physical examination findings.

Risk class I patients have:

●Age < 50 years

●The absence of these coexisting conditions:

•Neoplastic disease

•Heart failure

•Cerebrovascular disease

•Renal disease

•Liver disease

●The absence of these physical examination findings:

•Altered mental status

•Pulse ≥125/minute

•Respiratory rate ≥30/minute

•Systolic blood pressure <90 mmHg

•Temperature <35°C or ≥40°C

If age is 50 or less and none of the other factors exist, patients are in the lowest severity risk class I; if one or more risk factors are present, the evaluation of illness severity proceeds to Step 2 (table 1). The second step considers laboratory and radiographic information and stratifies the remaining patients into risk classes II, III, IV, or V based upon the total number of points assigned to each risk factor identified. A total point score is computed by adding the patient age in years (years minus 10 for females) and the points for each applicable risk factor. Total scores of 70 or under correspond to class II, 71 to 90 to class III, 91 to 130 to class IV, and over 130 to class V (calculator 1). In the derivation and validation of the PSI, any variables with missing information were considered normal.

Results — The PSI validation compared risk class-specific mortality rates [3]. All-cause 30-day mortality in risk classes I and II was low (ranging from 0.1 to 0.4 percent in class I and from 0.6 to 0.7 percent in class II); only four (0.3 percent) deaths in the three lowest risk classes were attributable to pneumonia. In contrast, 30-day mortality in risk class V was 27 percent (table 2).

In the initial validation study, increasing risk class was associated with a higher rate of subsequent hospitalization among those initially treated as outpatients and an increased frequency of intensive care unit (ICU) admission and longer length of stay for inpatients [3]. None of the 62 outpatients in risk classes I, II, or III who were subsequently hospitalized died, and only one was admitted to the ICU.

To estimate the impact of the rule, two admission strategies used projections based on patients in the Pneumonia PORT cohort [3]. The first strategy proposed outpatient therapy for risk class I or II patients, a brief stay in an observation unit or a short hospital admission for risk class III patients, and traditional inpatient care for patients in risk classes IV and V. This admission guideline would have reduced the rate of traditional inpatient admissions by 31 percent and recommended brief observation for another 19 percent. Among inpatients that would have been recommended for outpatient care under this approach, less than 1 percent died and 4.3 percent were admitted to an ICU. As noted above, risk class III patients can usually be managed as outpatients, as outpatients with home health care support, in an emergency department, or in a hospital observation unit. (See 'Approach to site of care' above.)

An additional margin of safety could use an amended strategy of traditional inpatient therapy for all risk class I, II, and III patients with hypoxemia. The latter approach would still have reduced traditional inpatient care by 26 percent; another 13 percent of inpatients would have been admitted for a brief observational stay. Under this second strategy, mortality among inpatients recommended for outpatient therapy was still less than 1 percent, but the proportion admitted to an ICU was reduced to 1.6 percent. Either admission strategy would have recommended inpatient therapy for five of the six Pneumonia PORT patients who died following an initial course of outpatient therapy.

Effectiveness and safety of the PSI as a decision aid — Three randomized trials examined the effectiveness and safety of using the PSI to guide the initial site of treatment for patients diagnosed with CAP in hospital emergency departments [19-22]. The following observations illustrate the range of findings in these trials:

●The CAPITAL trial randomized 19 Canadian hospital emergency departments to PSI implementation or usual care study arms. Among the 1743 patients enrolled with CAP, there was an 18 percentage point decrease in the admission of low-risk (risk class I, II, or III) patients in hospitals that implemented the PSI compared with hospitals in the control arm (49 versus 31 percent) [19]. In addition, implementation of the PSI had no negative effects on patient quality of life or the occurrence of adverse medical outcomes (ICU admission, mortality, hospital readmission, or complications).

●A prospective assessment of the safety of using the PSI rule to guide the initial site of treatment was performed in a randomized efficacy trial that compared outpatient care with hospitalization for CAP in 224 adults in risk class II or III (PSI scores ≤90 points) [21]. Patients were randomly assigned to outpatient care with oral levofloxacin or hospitalization with intravenous levofloxacin. Exclusion criteria included respiratory failure (PaO₂ <60 mmHg and/or saturation of ≤90 percent), complicated pleural effusion, unstable comorbid illness, severe social, cognitive, or psychiatric impairment, or inability to maintain oral intake.

The majority of patients in both groups had successful outcomes without complications (84 percent of outpatients compared with 81 percent of hospitalized patients). Treatment length, frequency of adverse drug events, and subsequent hospital admissions were similar in both groups. However, outpatients reported significantly greater satisfaction with care than inpatients (91 versus 79 percent).

●The emergency department community-acquired pneumonia (EDCAP) trial, conducted at 32 emergency departments in the United States, demonstrated that implementation of an admission guideline based upon the PSI safely increased the proportion of low-risk patients with CAP treated in the outpatient setting [20]. In this cluster-randomized trial, a practice guideline recommending outpatient treatment for 1901 low-risk patients (PSI risk classes I to III) without evidence of arterial oxygen desaturation, was implemented using randomly allocated low- (8 sites), moderate- (12 sites), and high-intensity (12 sites) guideline implementation strategies. Significantly greater proportions of low-risk patients treated at the moderate- and high-intensity implementation sites (61 and 62 percent, respectively) were treated as outpatients than at the low-intensity (38 percent) sites. No differences in safety outcomes (eg, mortality, medical complications, ICU admissions) occurred across the three intervention arms.

In this trial, factors independently associated with the hospitalization of low-risk patients were PSI risk classes II or III (versus class I), the presence of predefined medical or psychosocial contraindication to outpatient treatment, comorbid conditions not contained in the PSI, multilobar chest radiographic infiltrates, and home therapy with oxygen, glucocorticoids, or antibiotics before presentation [22]. Up to 20 percent of patients had no identified risk factor for hospitalization other than membership in risk class II or III, suggesting that physicians could safely discharge an even greater proportion of low-risk patients than observed in the moderate- or high-intensity arms of this trial.

The effectiveness and safety of using the PSI to guide the initial site of treatment decision are also supported by two quasi-experimental studies [23,24], one controlled observational study [25], and a meta-analysis of interventional studies [26]. All three of the individual studies show that use of the PSI to guide the site of treatment safely increases the proportion of low-risk patients treated as outpatients [23-25]. In a prospective observational study from France of patients with CAP seen in the emergency department, eight emergency departments used the PSI to determine site of treatment and eight did not [25]. A systematic review and meta-analysis identified six intervention studies guiding the initial site of treatment of low-risk patients with CAP [26]. All four studies that were clinical trials or prospective interventions with concurrent controls used the PSI to define low-risk patients. This study concluded that such interventions are effective, safe, and acceptable to patients.

Additional options for site of care — Although the initial site of treatment is often viewed dichotomously (home versus hospital), treatment intensity can be varied to meet the perceived clinical acuity needs of the patient. As an example, whereas traditional outpatient therapy with oral antibiotics may be appropriate for most low-risk patients with CAP, outpatient parenteral antibiotic therapy may be suitable for low-risk patients with intractable nausea or vomiting, poor gastrointestinal absorption, or for those who are less likely to adhere to oral therapy. In-home health care support may be an option for PSI risk class II and III patients who are older or have serious comorbid illnesses, abnormal vital signs, or borderline arterial oxygenation levels. Alternatively, such low-risk patients who may have other markers of illness severity and are at high risk for disease progression may benefit from the close monitoring and care available during a brief observational stay in the emergency department or hospital followed by discharge upon improvement. Although these types of treatment alternatives vary across medical institutions and insurance policies, each would presumably be more cost-effective than a traditional inpatient stay.

Limitations of using the PSI as a decision aid — A low risk of mortality is not the only dimension of decision-making for the initial site of treatment, as illustrated in a prospective observational study that examined the natural history of CAP in 586 Pneumonia Severity Index (PSI) risk class I and II patients who were hospitalized in Canada [27]. The following factors were independently associated with hospital admission:

●Female sex

●Presentation at a specific urban hospital

●Diminished premorbid functional status

●Comorbidities not contained in the PSI (eg, chronic obstructive pulmonary disease [COPD], asthma, heart disease, inflammatory bowel disease)

●Active substance abuse

●Psychiatric illness

●Respiratory rate ≥28 breaths/minute

●Shaking chills, shortness of breath, nausea, or diarrhea

Complications were reported in 111 (19 percent) of these low-risk patients. The most frequently reported complication was urinary catheter insertion (8.5 percent), which was unrelated to the underlying CAP; however, respiratory failure treated by mechanical ventilation occurred in 14 patients (2.4 percent) and empyema in 8 patients (1.4 percent). Five patients (0.9 percent) died. This underscores that some patients identified as low risk by the PSI have medical and/or psychosocial contraindications to outpatient therapy. The ability to maintain oral intake, history of substance abuse, cognitive impairment, severe neuromuscular disease, and patient functional status are other factors that may alter decision-making in favor of inpatient therapy. In addition, the rule should not be used for children, pregnant women, immunocompromised patients, and patients with hospital-acquired pneumonia, aspiration pneumonia, or suspected tuberculosis. Because this study did not evaluate clinically important patient exclusion criteria applied in previous studies [3], it could not directly assess the safety of using the PSI to guide the initial site of treatment. In particular, the failure to use arterial hypoxemia as an exclusion criteria (25 percent of their low-risk patients admitted had this condition) and the inclusion of patients meeting clear exclusions from a low-risk population eligible for outpatient treatment (eg, alcoholism, psychiatric illness, homelessness) overlook many potential safety features initially recommended by the PSI authors.

The PSI rule may oversimplify how physicians interpret some prediction variables. As an example, systolic blood pressures below 90 mmHg are considered abnormal, regardless of the actual value. A systolic blood pressure of 40 mmHg has a markedly different clinical implication than one of 85 mmHg, but the same points would be assigned to both measures.

A more practical limitation of using the PSI to guide decisions is its perceived complexity. Clinicians may consider calculating a score based upon 19 variables, placing patients into risk classes, and selecting an appropriate site for therapy to be too difficult or time consuming. However, this task can be streamlined using a calculator (calculator 1). Integration of the rule into a computerized system that automatically reminds physicians of the rule, computes a point score, and displays the recommended patient disposition could ease this task considerably.

All clinical decision aids, when used to guide the initial site of treatment, supplement rather than override physician judgment. Factors other than the predictors included in the rule may be important when making an admission decision for patients with CAP. In a study of patients evaluated in emergency departments with CAP who were found to be low risk (PSI classes I to III, no arterial oxygen desaturation or psychosocial contraindications to outpatient therapy), outcomes were compared for the 944 who received outpatient treatment and the 549 who were hospitalized [28]. Mortality at 30 days was higher for inpatients (2.6 versus 1.0 percent), suggesting that physician judgment regarding hospitalization was an appropriate adjunct to the risk stratification score. However, after propensity score matching for potential confounding factors contributing to site of treatment, mortality was the same for outpatients and inpatients, but outpatients had earlier return to work and activity.

Other applications of the PSI — The Pneumonia Severity Index (PSI) may help adjust for illness severity among patients with CAP, compare patient cohorts, or match clinical sites in a clinical trial. Comparisons between predicted and observed 30-day mortality within risk classes for patients presenting with CAP also could be a valuable continuous quality improvement tool for tracking hospital readmissions, mortality, physician performance, or practice patterns across facilities or health plans (table 2).

The CURB-65 and CRB-65 scores — The CURB-65 score is based upon five easily measurable factors from which its name is derived (table 3) [16]:

●Confusion (based upon a specific mental test or new disorientation to person, place, or time)

●Urea (blood urea nitrogen in the United States) >7 mmol/L (20 mg/dL)

●Respiratory rate ≥30 breaths/minute

●Blood pressure (BP; systolic <90 mmHg or diastolic ≤60 mmHg)

●Age ≥65 years

Among the 718 patients (mean age 64) in the derivation cohort of CURB-65, 30-day mortality was 0.7, 2.1, 9.2, 14.5, and 40 percent for 0, 1, 2, 3, or 4 factors; only a small number of patients had 5 factors. Similar findings were noted in the separate validation cohort. When the two cohorts were combined, the mortality rate was 0.6 percent with 0 factors (1 of 173 patients) and 1.7 percent with 1 factor (4 of 241 patients). Although based on a relatively small sample size, the authors suggested that patients with a CURB-65 score of 0 to 1 could be treated as outpatients, those with a score of 2 should be admitted to the hospital, and those with a score of 3 or more should be assessed for ICU care, particularly if the score was 4 or 5 (calculator 2). In a large independent study of 3181 patients with CAP, mortality was 3 percent among 901 patients with a CURB-65 score of 1 [29]. The mortality rate within this CURB-65 risk group is like that observed for PSI class III patients; although some of these patients can be safely managed as outpatients, others may warrant brief emergency department or hospital observation or brief inpatient hospitalization. We generally favor hospital admission for patients with a CURB-65 score of 1, although patients with a score of 1 due to being ≥65 years of age who do not have major comorbidities may not require admission. The CURB-65 is an alternative to the PSI, but its effectiveness and safety in guiding the initial site of treatment has never been empirically assessed.

A simplified version (CRB-65), which does not require testing for blood urea nitrogen, is sometimes used for decision-making in primary care practitioner's offices [30]. In this version, admission to the hospital is recommended if one or more points are present. A study of 670 patients from Germany found good utility from the CRB-65 score [30]. However, further effectiveness and safety studies are needed for all CURB score variants before they can be recommended as decision aids.

Comparison to PSI and PSI-high risk (HR) — In a prospective study of 3181 adults with CAP, investigators compared the performance of the Pneumonia Severity Index (PSI), CURB-65, and CURB in predicting 30-day mortality [29]. Overall, the PSI classified 68 percent of the patients as low risk, more than the CURB (51 percent) or the CURB-65 (61 percent). Although aggregate 30-day mortality rates were slightly lower for low-risk patients identified by the PSI, the clinical relevance of the small differences is unclear.

In an analysis of patient-level data from six international pneumonia cohorts, a modified version of the PSI was developed to more accurately identify high-risk patients (PSI-HR) using the original prognostic weights of all PSI variables and modifying the risk thresholds to define its six risk classes [31]; the prognostic accuracy of PSI-HR and CURB-65 were compared using 30-day mortality as the outcome. In 13,874 adults with CAP, PSI-HR classified a significantly larger proportions of patients as low-risk (55.2 versus 52.7 percent) and high-risk (5.6 versus 4.7 percent) than CURB-65. For PSI-HR versus CURB-65, aggregate mortality was significantly lower in low-risk patients (1.6 versus 2.2 percent) and non-significantly higher in high-risk patients (36.5 versus 32.2 percent).

Future comparisons of PSI and CURB based prediction rules for guiding the initial site of treatment are needed, as are comparisons of the prognostic performance of the PSI-HR and other scores to identify high-risk patients with severe CAP.

Scores to identify severe CAP — Several scoring systems exist to identify severe CAP, including the severe CAP (SCAP) score, SMART-COP, and the Infectious Diseases Society of America (IDSA)/American Thoracic Society (ATS) major and minor severity criteria. As noted above, the presence of either one of the two IDSA/ATS major criteria (septic shock requiring vasopressor support and requirement for mechanical ventilation) dictate urgent transfer to a critical care site (ICU or emergency department) (see 'Approach to site of care' above). Although the SCAP score, SMART-COP, and the IDSA/ATS minor criteria require further study before they can be applied as decision aids to guide the intensity of inpatient treatment, it is reasonable to consider the IDSA/ATS minor criteria when selecting the acuity level of care for hospitalized patients who do not meet other criteria for ICU admission.

SCAP — The severe community-acquired pneumonia (SCAP) score seeks to predict in-hospital mortality, need for mechanical ventilation, and risk for septic shock [17]. In a multivariate analysis in a derivation cohort, the following criteria were independently associated with severe CAP:

●Major criteria:

•Arterial pH <7.30 – 13 points

•Systolic blood pressure <90 mmHg – 11 points

●Minor criteria:

•Respiratory rate >30 breaths/minute – 9 points

•PaO₂/FiO₂ <250 mmHg – 6 points

•Blood urea nitrogen >30 mg/dL (10.7 mmol/L) – 5 points

•Altered mental status – 5 points

•Age ≥80 years – 5 points

•Multilobar/bilateral infiltrates on radiograph – 5 points

In the original validation study, a SCAP score ≥10 (which represents at least one major criterion or at least two minor criteria) was superior to the PSI and CURB-65 scores for predicting evolution toward severe CAP [17]. In a subsequent validation study, the SCAP score was at least as accurate as the other prognostic scoring systems for predicting intensive care unit admission, need for mechanical ventilation, progression to severe sepsis, and treatment failure [32]. Further studies are needed before the SCAP score should be used to inform site of treatment (home versus hospital) or ICU admission decisions.

SMART-COP — One study evaluated a set of clinical criteria to predict need for intensive care in 882 patients with an episode of CAP [33]. More than 75 percent of patients were over 50 years old. The following factors were associated with the need for intensive respiratory or vasopressor support: low systolic blood pressure (less than 90 mmHg), poor oxygenation, low arterial pH, multilobar pneumonia, hypoalbuminemia, high respiratory rate, tachycardia, and confusion. The investigators designed a scoring system (called SMART-COP) in which each factor led to the accrual of one point, except for low systolic blood pressure, poor oxygenation, and low arterial pH, each of which contributed two points [33]. A score of ≥3 points identified 92 percent of patients who received intensive respiratory or vasopressor support. The accuracy of this system was also high in five validation databases.

In a separate study of CAP in 335 young adults (<50 years of age), SMART-COP was superior to CURB-65 and PSI for predicting the need for mechanical ventilation and/or inotropic support but failed to identify 15 percent of patients who required these measures [34]. Thus, the sensitivity of SMART-COP appears to be substantially reduced in individuals younger than age 50. Similarly, its use to guide the intensity of inpatient treatment should await future studies of its application as a decision aid.

IDSA/ATS severity criteria — The Infectious Diseases Society of America (IDSA)/American Thoracic Society (ATS) consensus guidelines define two major criteria for requiring admission to an ICU: septic shock requiring vasopressor support and requirement for mechanical ventilation [35]. The guidelines also recommend ICU care for patients with at least three of the following minor IDSA/ATS severity criteria:

●Altered mental status

●Hypotension requiring fluid support

●Temperature <36°C (96.8°F)

●Respiratory rate ≥30 breaths/minute

●PaO₂/FiO₂ ratio ≤250

●Blood urea nitrogen ≥20 mg/dL (blood urea 7 mmol/L)

●Leukocyte count <4000 cells/microL

●Platelet count <100,000/mL

●Multilobar infiltrates

A validation study of over 2000 patients with CAP confirmed the validity of the major criteria [36,37]. In a subsequent validation study of 1062 patients with CAP who did not have the major criteria or a contraindication to ICU admission, the minor criteria were superior to other prognostic scoring systems (including the PSI, CURB-65 score, and SCAP) for predicting the need for mechanical ventilation, vasopressor support, and ICU admission [38]. The minor criteria were equivalent to the SMART-COP scoring system for predicting these outcomes.

PRACTICE GUIDELINES — Practice guidelines for the management of patients with CAP all make recommendations for the initial site of treatment and/or the acuity of inpatient care for patients diagnosed with CAP. Probably the most widely recognized medical practice guidelines are those published by the British Thoracic Society (BTS) [39], the National Institute for Health and Care Excellence (NICE) [40], and the consensus guidelines from the Infectious Diseases Society of America (IDSA) and the American Thoracic Society (ATS) [35].

Site of treatment — Although all three guidelines recommend the use of prediction rules for prognosis to guide the initial site of treatment decision, the recommendations differ on which ones to use. The NICE guidelines recommend using the CURB-65 (primarily in hospital) or CRB-65 (primarily in a community setting), with outpatient management for a score of zero and hospital referral for a score of one or two [39-41]. The IDSA/ATS guidelines recommend preferentially using the PSI over CURB-65 to guide site of treatment as a supplement to physician judgment [35]. This recommendation is based on the superior accuracy of the PSI and its empirically established safety and effectiveness in guiding clinical decision-making based upon multiple clinical trials and quasi-experimental studies. Although we prefer the PSI for the same reasons, CURB-65 is an alternative that can be used when a less complex scoring system is desired, but its effectiveness and safety have not been assessed in any clinical trial. Our recommendations about site of care are presented above. (See 'Approach to site of care' above.)

Admission to intensive care — The BTS guidelines recommend using the CURB-65 to guide the acuity of inpatient care, with a critical care unit (high dependency or intensive care unit [ICU]) suggested for patients with scores of four or five [39,40]. The NICE guidelines state that ICU admission should be considered for patients with a CURB-65 score ≥3 [40]. As stated above, the IDSA/ATS consensus guidelines recommend major and minor clinical criteria to guide admission to intensive care [35]. We recommend using the IDSA/ATS major and minor criteria for guiding the acuity levels of inpatient care due to their widespread validation. However, future research is required to assess the effectiveness and safety of using prediction rules to guide the acuity of inpatient care.

IMPORTANCE OF OXYGEN SATURATION — Patients with blood oxygen saturations <90 percent are usually hospitalized; this criterion was a recommended trigger for hospital admission in the original study of the Pneumonia Severity Index (PSI) and has been used as an independent criterion for admission in most trials of the PSI to guide the initial site of treatment [3,20,21]. A subsequent population-based cohort study showed that oxygen saturations <92 percent may represent a more appropriate threshold to guide the initial site of treatment of patients with CAP [42]. Among 2923 patients with pneumonia who presented to the emergency department but who were discharged to be treated as outpatients, those with oxygen saturations <90 percent had significantly higher rates of 30-day mortality (6 versus 1 percent), hospitalization (18 versus 7 percent), and the composite endpoint of either 30-day mortality or hospitalization (21 versus 8 percent) than patients with oxygen saturations >90 percent. Blood oxygen saturation <90 percent was independently associated with 30-day mortality or hospitalization (adjusted odds ratio 1.7, 95% CI 1.1-2.8). If the oxygen saturation threshold was increased to 92 percent, there was no association with adverse events; such a change in threshold would have entailed one additional hospitalization for every 14 patients discharged. These findings suggest that using a blood oxygen saturation of <92 percent would be a safer threshold for hospital admission than a threshold of <90 percent. These thresholds for arterial hypoxemia are most applicable to patients without underlying pulmonary disease who have evidence of new arterial desaturation on room air associated with their pneumonia. Home oximeters are now available, not costly, and have good fidelity, allowing closer monitoring of those treated outside the hospital.

NURSING HOME RESIDENTS — Lower respiratory tract infections are common among residents of nursing homes and one of the most frequent reasons for transferring residents to a hospital [43,44]. However, hospitalization may be associated with a reduction in quality of life and decline in functional status [45]. For these reasons, investigators in Canada designed and evaluated a clinical pathway for treating residents with respiratory tract infections, including pneumonia, in the nursing home [46]. Twenty-two nursing homes were randomly assigned to use of a clinical pathway or usual care in treatment of lower respiratory infections, and a total of 680 nursing home residents ≥65 years were enrolled in the study.

The presence of two or more symptoms or signs of lower respiratory tract infection (eg, new or increased cough, new or increased sputum production, temperature >38ºC, pleuritic chest pain, or new or increased abnormal findings on chest examination) and a new or increased infiltrate on chest radiograph defined pneumonia. Study nurses assessed patients and provided care in the nursing home in those who met the following criteria:

●Able to eat and drink

●Pulse ≤100 beats per minute

●Respiratory rate ≤30 per minute

●Systolic blood pressure ≥90 mmHg or a decrease of ≤20 mmHg from baseline if <100 mmHg

●Oxygen saturation ≥92 percent or if the resident had chronic obstructive pulmonary disease ≥90 percent

The following other observations are important:

●Patients treated by the clinical pathway were less likely to be hospitalized (10 versus 22 percent hospitalized in the usual care group).

●There was no difference in mortality rate (8 versus 9 percent), health-related quality of life, or functional status between the clinical pathway versus usual care groups, respectively.

●The clinical pathway resulted in an overall cost savings of USD $1016 per resident treated.

This study is limited in that only 36 percent of the study population had radiographically confirmed pneumonia. Also, enrolled patients were from nursing homes with ≥100 beds, decreasing the generalizability to patients from smaller nursing homes. Finally, although this intervention was implemented in a single-payer health care system (Canada), it may be more difficult to implement in a multiple-payer system (eg, the United States).

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Community-acquired pneumonia in adults".)

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5^th to 6^th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10^th to 12^th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or email these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Beyond the Basics topic (see "Patient education: Pneumonia in adults (Beyond the Basics)")

SUMMARY AND RECOMMENDATIONS

●Hospital admission rates – Hospital admission rates in the United States for adults with community-acquired pneumonia (CAP) vary widely and are often not directly related to local disease severity clinicians often use inconsistent criteria when making the initial decision about the appropriate site of care. In addition, clinicians often overestimate patient risk of short-term mortality in low-risk patients, resulting in missed outpatient care opportunities and unneeded cost. (See 'Introduction' above.)

●Determining need for hospital admission

•Severity of illness is the most critical factor in triaging a patient, but other factors should be taken into account (algorithm 1). These include the ability to maintain oral intake, likelihood of medication adherence, history of active substance use disorder, mental illness, cognitive or functional impairment, living or social circumstances (eg, homelessness, residence far enough from a health care facility that precludes timely return to care in the event of clinical worsening), and clues about select microbial pathogens. (See 'Approach to site of care' above.)

•If the patient has septic shock or respiratory failure, urgently transfer to a hospital critical care site (intensive care unit [ICU] or emergency department). (See 'Admission to intensive care' above.)

•If there are practical concerns about outpatient management (such as those outlined above) or evidence of arterial hypoxemia (oxygen saturation is <92 percent on room air), observation with reassessment or hospital admission is warranted. (See 'Importance of oxygen saturation' above.)

●Clinical prediction rules for hospital admission – If the decision for site of care isn't set by the factors above, use a validated clinical prediction rule for prognosis to supplement physician judgment. (See 'Approach to site of care' above.)

•Among the available prediction rules, we prefer the Pneumonia Severity Index (PSI) because it is more accurate prognostically than CURB-65 and its safety and effectiveness in guiding clinical decision-making have been empirically confirmed in multiple studies (calculator 1). (See 'Pneumonia Severity Index' above.)

-Patients who are PSI class I or II can be managed as outpatients. Some PSI class II patients may benefit from in-home health care support, also termed "hospital-at-home" (eg, a visiting nurse, intravenous fluids, intravenous antibiotics).

-Patients who are PSI class III can usually be managed as outpatients, as outpatients with in-home healthcare support, or with a brief observational stay in an emergency department or inpatient observation unit. The choice of site of care will depend on each patient's clinical status and other circumstances.

-Patients who are PSI class IV or V should be admitted to the hospital, and ICU admission should be considered.

•The CURB-65 score (calculator 2) is a less complex alternative for prognosis, but its safety and effectiveness in guiding the initial site of treatment have not been empirically assessed.

-Patients with a CURB-65 score of 0 can be managed as outpatients.

-Patients with a CURB-65 score of 1 or 2 should generally be admitted to a general medical ward, although patients who receive a score of 1 because they are ≥65 years of age who do not have major comorbidities do not necessarily require hospital admission.

-Patients with a CURB-65 score of 3 to 5 should be admitted to the hospital, and ICU admission should be considered.

●IDSA/ATS criteria for ICU admission – For hospitalized patients, other additional markers of illness severity support higher acuity levels of inpatient care. Patients with ≥3 Infectious Diseases Society of America/American Thoracic Society minor severity criteria often warrant admission to a higher acuity level of inpatient care, including the ICU. (See 'IDSA/ATS severity criteria' above.)

ACKNOWLEDGMENT — UpToDate gratefully acknowledges John G Bartlett, MD (deceased), who contributed as Section Editor on earlier versions of this topic and was a founding Editor-in-Chief for UpToDate in Infectious Diseases.