Community-acquired pneumonia (CAP) continues to generate a large amount of interest, both for the clinician and the researcher. It is a very frequent diagnosis and the leading infection-related cause of death in most developed countries.1

Although CAP is a relatively common infection, there are wide disparities in its management, including the class of antibiotics chosen, the duration of therapy and the role of adjunctive therapy such as corticosteroids. In this review, we assess the evidence for the approaches to some of these clinical questions regarding CAP management. We agree with the Australian antibiotic guidelines2 regarding recommended antibiotics. Therefore, we do not specifically consider the question of the most appropriate class of antibiotics for treating patients with CAP — the Box summarises the antibiotics commonly used in Australia.

We used a PubMed search for original and review articles from 2005 to 2017, and reviewed specialist society publications and guidelines from Australia and overseas, to formulate an evidence-based overview of the topic as applied to clinical practice.

Are we overdiagnosing CAP?

Although it may seem self-evident, an essential question in the management of patients with CAP is whether the diagnosis is in fact correct. CAP can present in variable ways, some of which are similar to other conditions such as acute bronchitis, viral respiratory tract infections and cardiac failure. Patients with dementia, who are more likely to develop CAP, may not be able to give a reliable description of symptoms.3 Patients may present with two or more conditions at once, confusing the diagnostic process.3 This may occur as a coincidence or alternatively be due to a cause–effect relationship between them. Examples of the latter include that a chest infection can precipitate either an exacerbation of cardiac failure or an acute coronary syndrome.4 In addition, particularly in the era of the 4-hour National Emergency Access Target, staff members in the emergency department (ED) are under greater pressure to move patients out of the ED and thus may need to change the focus of their assessment to “does this patient need admission?” rather than “what is the correct diagnosis?”.

From clinical studies of CAP performed in Australia, of all the patients screened for inclusion on the basis of being given the label of CAP in the ED, a large proportion are subsequently excluded from the study because their chest x-ray is not consistent with CAP.5,6 This issue is not limited to Australia, with international studies showing that chest x-rays reported by treating clinicians as being consistent with CAP are not confirmed as being so by a radiologist in 20–50% of cases.7–11

There are several downsides to excessive diagnosis of CAP. The most obvious is the use of unnecessary antibiotics in patients who have conditions that do not require antibiotics such as viral respiratory infections or cardiac failure. This has the potential to add to the problem of antibiotic resistance as well as putting the patient at risk of antibiotic-related complications such as Clostridium difficile-associated diarrhoea. A further issue, particularly when cultures are not performed in patients initially labelled as having CAP, is the potential delay in diagnosis and inappropriate antibiotic therapy of those patients whose true diagnosis is something more serious, such as sepsis, infective endocarditis or pulmonary embolism. Some of these misdiagnosed patients can have their admission prolonged by many days due to the non-performance of blood cultures. We believe that the diagnostic uncertainty for admitted patients initially given the diagnosis of CAP means that recommendations that discourage the performance of blood cultures in CAP patients are inappropriate.12–15

Duration of antibiotic therapy

The optimal duration of antimicrobial therapy for CAP is another area of controversy. The tendency in hospitals appears to be to overtreat rather than undertreat, often with a long oral tail.16–18 Whether this is a case of believing that “more is better” or due to the disparity between the time to clinical resolution compared with microbiological resolution, the excessive prescription of antibiotics puts the patient at greater risk of side effects and colonisation with resistant organisms, including nasopharyngeal carriage of penicillin-resistant Streptococcus pneumoniae.19,20 Ecologically, the prescription of antibiotics for respiratory infection contributes to a rise in resistance in the community.21

Should the physician turn to national guidelines for advice on duration and choice of antibiotic (Box); the Australian Therapeutic guidelines: antibiotic recommend 7 days of total therapy for moderate and most cases of severe pneumonia,2 as does the British Thoracic Society,22 while the United Kingdom NICE guidelines suggest 5 days for mild CAP and 7–10 days for moderate to severe CAP.23 However, the Infectious Diseases Society of America (IDSA) supports a 5-day treatment for inpatient CAP, provided the patient is afebrile and clinically improving.24 So, with all this variation, which is correct?

There is agreement that a 7-day course of an antibiotic is effective for most cases of CAP, and this is relatively non-controversial, albeit adhered to poorly.25 There is increasing evidence, however, that shorter courses of 5 or even 3 days’ therapy may be just as effective. Overseas literature provides support for short course therapy with azithromycin, including as little as a single dose.26 This likely relates to the high tissue penetration and persistence of adequate tissue levels of this macrolide for some days following administration.27 A multicentre randomised clinical trial evaluating the safety of the IDSA recommendations found that a 5-day course of therapy is safe and effective, although most patients received quinolone antibiotics, a class of antibiotic rarely used in Australia for treating CAP.28 Regarding the β-lactam therapy that would be more likely prescribed in the Australian setting, a 3-day course of intravenous (IV) amoxycillin monotherapy has been shown to be as effective as 3 days of IV amoxycillin followed by 5 days of oral amoxycillin in adult patients who were improving at 72 hours.29 Two previous studies reached a similar conclusion in paediatric populations.30,31

Given the accumulating evidence, we suggest that a 5-day course of antibiotics should be effective in most cases of uncomplicated CAP, even though complete symptom resolution is unlikely to have occurred at this time point. For patients on IV therapy who are clinically improving at 72 hours, a switch to oral therapy is appropriate, but clinicians should keep in mind that the oral antibiotics should complete the 5-day total course and not add another 5 days to what has already been prescribed. If improvement has been rapid in the first 72 hours, it would be reasonable to cease all therapy at 3 days, provided close follow-up is available.

Some international studies have suggested that bundles of care for patients with CAP, which include antibiotic administration within 4 to 8 hours of presentation, may lead to better patient outcomes.32–34 However, it is not clear that this would provide benefits in the Australian setting. In relation to the United States studies,33,34 this finding may reflect past differences in the US health system, where antibiotics may not have been given until the patient was seen by their attending physician, potentially leading to delays in therapy. The US recommendations have now changed to recommend commencement of antibiotics while the patient is in the ED.24 This is already the norm in Australia.

Other studies35,36 have suggested that increases in mortality in patients with CAP may be due to an atypical presentation which leads to a delay in diagnosis, rather than being associated with a delay in antibiotic administration. When this was taken into account in one study, the association between a delay in antibiotic administration beyond 4 hours and increased mortality was not statistically significant.35

Potential cardiac side effects of newer macrolide antibiotics

A 2012 study reported an excess of both cardiovascular and all-cause deaths in patients with pneumonia treated with a 5-day course of azithromycin compared with those treated with other antimicrobials, potentially related to its ability to prolong the QT interval.37 As a result, in 2013, the US Food and Drug Administration issued a warning regarding prescription of azithromycin for CAP, even though that study had a number of limitations, including its non-randomised nature and outpatient study population.

However, the case was far from closed, and results from other retrospective studies reached the opposite conclusion. Mortensen and colleagues studied older patients with CAP and found that those treated with macrolides had a lower rate of mortality, in spite of a small rise in rates of myocardial infarction “consistent with a net benefit”.38 This conclusion was shared by Cheng and colleagues in their 2015 meta-analysis.39 In 2016, a Canadian population-based retrospective cohort study involving about one million adults aged over 65 years found no increase in rates of cardiac arrhythmias at 30 days, in addition to lower all-cause mortality, in patients treated with a macrolide antibiotic.40

Given the evidence that the benefit of using macrolide therapy outweighs potential cardiac risk, we support recommendations to use a macrolide in place of doxycycline for atypical cover when the latter cannot be used, and the use of azithromycin in combination therapy for severe hospitalised CAP, such as that requiring management in an intensive care unit (ICU). We also point out the excellent oral bioavailability of oral azithromycin,27 and recommend its use in preference to the IV formulation in patients for whom oral therapy is tolerated and expected to be absorbed.

The link between CAP and cardiovascular disease

In recent years, evidence has emerged regarding the role of inflammatory conditions in the development of cardiovascular disease such as myocardial infarctions and strokes.41 It is postulated that inflammation, especially when persistent, may have an effect on vascular plaques, making them more unstable or prone to acute occlusion.42,43 Various infections including CAP, influenza and human immunodeficiency virus, as well as other sources of chronic inflammation such as rheumatoid arthritis, have all been shown to be associated with higher rates of acute cardiovascular disease and deaths.4,44–51

In a large study, in the 30 days following an episode of CAP requiring inpatient care, incidence of worsening heart failure, cardiac arrhythmia and acute myocardial infarction were 21%, 10% and 3% respectively.4 However, it is important to note that the problem does not end after 30 days. There is a measurably higher rate of cardiovascular deaths in the following few years, when patients admitted with CAP are compared with matched cohorts admitted with non-infection-related conditions. The rate increases most in older patients (aged over 40 years) and those with greater number of cardiovascular risk factors.52

The mechanism of this increase in cardiovascular complications during and after the CAP episode appears to be multifactorial. Inflammation is a pro-thrombotic state; myocardial inflammation and damage may occur, potentially in response to NADPH oxidase 2 upregulation; cardiac strain may be present in the setting of increased sympathetic nervous system activity with relative hypoxia caused by the lung consolidation; increased fluid and sodium loading associated with some IV antibiotic may worsen fluid overload problems in some cardiac failure patients; and QT interval prolongation with the use of other antibiotics may contribute to arrhythmic potential.46,47,53

What remains to be seen is whether we can act on this in a useful way. It is notable that the vast majority of patients who die from CAP are very old with multiple comorbidities, for whom death may be an expected terminal event. While acutely addressing cardiac risk factors with, for example, the addition of anti-platelet agents like aspirin or cholesterol-lowering statin therapy has not yet been shown to alter mortality in the acute setting,54 it would appear prudent to assess whether such treatments are indicated in patients admitted with CAP, especially if they are aged over 40 years.52

The role of corticosteroids in the management of CAP

Given that the inflammatory state during and after an episode of CAP appears to have an important role in contributing to both morbidity and mortality,4,44–47 there has been interest in the role of inflammatory modulators such as corticosteroids as adjunctive CAP therapy. Levels of cytokines vary with severity of CAP and highest levels of the pro-inflammatory cytokine interleukin (IL)-6 and the anti-inflammatory cytokine IL-10 are associated with higher chance of dying from severe CAP.55 Glucocorticoids reduce the levels of such cytokines,56 and thus are theoretically attractive as a means to reduce CAP mortality.

There have been a number of attempts to address the question about whether this theoretical benefit may be true. Individual studies have varied in terms of the severity of the CAP studied, the choice of corticosteroid used, the route by which it was given, its dose and duration, and the outcomes measured. Results have been mixed, and several attempts at performing meta-analyses on these studies — with all the expected problems associated with attempting to combine such a heterogeneous collection of methodologies — have shown marginal benefits in terms of mortality, particularly in patients with the most severe CAP managed in the ICU, as well as a shorter time to becoming afebrile.57–60 These small benefits need to be weighed against the potential downside of high-dose corticosteroids, both in terms of potential side effects like immune suppression and also the fact that outcomes may have been worse in patients whose infection was caused by an influenza virus or Aspergillus.61,62

Thus, the potential role of corticosteroids as adjunctive therapy in CAP appears to be very limited. They could be considered in patients with CAP severe enough to require management in the ICU, but caution should be taken until the aetiology is known, particularly during influenza season. Their use should also be very carefully considered in patients at higher risk from corticosteroid complications, such as the immunocompromised, women who are pregnant, patients with recent gastrointestinal haemorrhages, and patients at greater risk of neuropsychiatric problems.59 The possible shortened time to defervescence is not sufficiently clinically useful to justify the potential harm from such therapy.

Conclusion

In this era of burgeoning antibiotic resistance, the treatment of CAP is an area where we have the potential to reduce antibiotic consumption. We are diagnosing it too often and treating it for too long. Most non-ICU patients with CAP could be treated for 3–5 days in total.

CAP is a common cause of death, both in the short term and also in the subsequent few years, and many of these deaths appear to be cardiovascular related. Although most deaths from CAP occur in very old people with multiple comorbidities — and so may not easily be prevented — the management of a patient with CAP should be seen as an opportunity to address and treat cardiac risk factors when they are present.