This bread-and-butter ICU topic has appeared in the exams many times. Relevant questions have included the following:
- Question 2 from the first paper of 2018 asked about the factors which influence decisions regarding ICU admission.
- Question 21 from the second paper of 2017 was about non-resolving CAP
- Question 5 from the first paper of 2016 was all about the use of steroids in CAP.
- Question 18 from the first paper of 2012 discusses antibiotic management and prognostic issues.
- Question 7 from the second paper of 2012 looks indepth at the causes of treatment failure, and the investigations for treatment-refractory pneumonia.
- Question 26 from the first paper of 2006 asks again about specific features of history which would be important
- Questions 2a, 2b, 2c and 2d from the second paper of 2003 are a thorough exploration of CAP in general.
The most important primary sources for the information summarised here would have to include the following:
- 2007 Infectious Disease Sociaty of America Guidelines
- 2009 British Thoracic Society Guidelines
- 2014 National Institute for Health and Care Excellence (NICE) Pneumonia Guidelines
They are all pretty much the same, making a series of similar recommendations. ASID is the local body in charge of this, and they have yet to issue a position statement on Australian pneumonia. We must assume that one infected lung is much like another, and rely on foreign Thoracic Societies.
In Question 18 from the first paper of 2012, the college asked for "features which suggest a poor prognosis".
As their model answer, they offered the candidates a cut-and-pasted table from the 2007 ISDA Guidelines. I reproduce it below:
There are several alternative scoring systems:
- PSI Pneumonia Severity Index (Flanders et al, 1999)
- CURB65 score (Lim et al, 2003), recommended by the 2014 NICE guidelines
- SMART-COP (Charles et al, 2008)
- IDSA/ATS criteria (Mandell et al, 2007)
Which of these tools is best? According to a 2010 meta-analysis by Chalmers et al, they are all much the same. Niederman (2009), in a narrative review, suggests that these tools mainly differ according to the intended application. The old PSI and the NICE guideline-recommended CURB65 score are both very good at identifying good for low-morality patients and there therefore better in the ED and the GP clinic. The IDSA/ATS guidelines and the SMART‐COP tool are probably better at identifying the patients at need of ICU care. For the purposes of completeness, the SMART-COP tool is included below:
For completeness, here's the CURB65 score, which is calculated by giving 1 point for each of the following prognostic features:
- confusion (abbreviated Mental Test score 8 or less, or new disorientation in person, place or time
- raised blood urea nitrogen (over 7 mmol/litre)
- raised respiratory rate (30 breaths per minute or more)
- low blood pressure (diastolic 60 mmHg or less, or systolic less than 90 mmHg)
- age 65 years or more.
Patients are stratified for risk of death as follows:
- 0 or 1: low risk (less than 3% mortality risk)
- 2: intermediate risk (3‑15% mortality risk)
- 3 to 5: high risk (more than 15% mortality risk).
Why is risk stratification important? Prragmatically, it is because Question 2 from the first paper of 2018 asked about the factors which influence the decision to admit to ICU, vs. send home or to the ward. It makes greatest sense that ICU admission should be considered for the high risk patients, i.e. those at greatest risk of mortality. One might also add that ICU admission would be inappropriate for patients in whom mortality is expected to be inevitable regardless of treatment, and that should probably also factor into the decision.
Historical features of specific interest
Question 26 from the first paper of 2006 and Question 2a from the second paper of 2003 request that the candidate take a rich full history from the patient and their family, focusing on those things which might have some importance in community-acquired pneumonia. The two questions are essentially identical. The "model" college answer has been reworked into a list of important historical features, which is reproduced below:
Patient's medical history of prognostic importance
- heart disease, including CCF and pulmonary hypertension
- exercise tolerance
- immunesuppression, eg. corticosteroids
- vaccination history
- history of exposure to TB, country of origin
- end-stage organ failure, eg. cirrhosis or dialysis-dependent renal failure
Recent history of aetiological importance
- swallowing difficulty, history of stroke
- alcohol binges
- recent travel
- recent gardening (Legionella)
- recent bushwalking
- exposure to birds or pets
- recent chemotherapy
- recent hospital stay
- recent antibiotic use
- rapidity of onset
- associated symptoms eg. copious amounts of sputum, flu-like symptoms, hemoptysis, pleuritic chest pain, and so forth
Features which help narrow the organism
- Recent hospital stay, nursing home residence (community acquired vs. nosocomial: determines the likelihood og Gram negative organisms, and influences antibiotic decisions)
- Recent travel to regions of known Legionella epidemics
- Exposure to birds (Chlamydia psittacii)
- Exposure to groups with known drug-resistant pneumococcus, or similar
- Risk factors for Pseudomonas infection
- Structural lung disease, eg. bronchiectasis
- Use of corticosteroids
- Courses of broad-spectrum antibiotics
- Specific immunosuppression:
- Splenectomy (encapsulated organisms: S.pneumoniae, H.influenzae)
- HIV (Pneumocystis, tuberculosis)
- Neutropenia (fungi)
The cavernous innards of the BTS guidelines statement contain the very useful "Box 1" (page iii16) which describes "clinical features reported to be more common with specific pathogens". The content of this box is reproduced below to speed revision:
Clinical features of specific pathogens:
- Streptococcus pneumoniae: increasing age, comorbidity, acute onset, high fever and pleuritic chest pain.
- Bacteraemic S. pneumoniae: female sex, excess alcohol, diabetes mellitus, chronic obstructive pulmonary disease, dry cough.
- Legionella pneumophila: younger patients, smokers, absence of comorbidity, diarrhoea, neurological symptoms, more severe infection and evidence of multisystem involvement (eg, abnormal liver function tests, elevated serum creatine kinase).
- Mycoplasma pneumoniae: younger patients, prior antibiotics, less multisystem involvement.
- Chlamydophila pneumoniae: longer duration of symptoms before hospital admission, headache.
- Coxiella burnetii: males, dry cough, high fever.
Common pathogens in community-acquired pneumonia
In Question 18 from the first paper of 2012, the college again quotes directly from the ISDA guidelines.
The most common microbes are:
- Streptococcus pneumoniae
- Mycoplasma pneumoniae
- Haemophilus influenzae
- Chlamidophyla pneumonia
- Respiratory viruses
Investigations for community-acquired pneumonia
In the model answer to Question 2b from the second paper of 2003, the college lists a series of investigations which they would expect their candidate to mention. Of the following list of tests, the normal font identifies the college canon, and the italicised ones were added by the deranged author.
- Arterial blood gas
- Blood cultures
- FBC (for leucocyte morphology rather than their number)
- EUC (to anticipate renal clearance problems)
- LFTs (to anticipate hepatic clearance problems)
- Blood culture
- Sputum culture
- Pleural fluid culture
- Legionella urinary antigen
- Pneumococcal urinary antigen
- Bronchoscopic lavage specimen for culture
- Respiratory viral swabs (NAT)
- Inflammatory markers, eg. CRP and procalcitonin
This list of diagnostic tests was assembled from the 2007 IDSA guidelines statement. There are arguments against the routine use of blood and sputum cultures, given that their yield is poor and they do not tend to influence management. Similarly, urinary antigen tests do not tend to change your antibiotic decisions for the majority of patients However, the IDSA recognises that some patients are sick enough to merit an extended spectrum of tests. Table 5 from their publication lists the indications for such extended testing. ICU admission is at the top of that list. Ergo, any patient asked about in the CICM exam is deserving of extra attention by default.
The evidence and utility of these investigations can be examined in some greater detail:
- CXR : according to the IDSA and BTS, "a demonstrable infiltrate by chest radiograph or other imaging technique... is required for the diagnosis of pneumonia".
- ABG: recommended by the BTS for all patients receiving "emergency oxygen". In studies of random mixes of patients, ABG increased the chances of detecting clinically significant hypoxia. It is frightening to think that doctors need an ABG to accurately identify patients with clinically significant hypoxia, but in reality the blood gas result may sway one's decision regarding oxygen supplementation and intubation (eg. a patient who is tachypnoeic, and becoming hypercapnoeic due to fatigue).
- Blood cultures: all the guidelines recommend that patients with pneumonia of ICU-level severity should get cultured. The specific caveat is that cultures should be collected "whenever the result is likely to change individual antibiotic management". The de-escalation of broad spectrum antibiotics would represent such a change.
- Sputum culture is neither sensitive nor specific, but the BTS and IDSA recommend that we send purulent samples. Of particular interest are tracheal aspirate and BAL samples (as they might actually be representative of the infectious agent)
- Pleural fluid culture is valuable if the patient already has empyema, or if one is suspected. This becomes of more interest in the pneumonia which fails to resolve.
- Pneumococcal urinary antigen (specifically pneumococcal capsular polysaccharide antigen) has broad popularity and multiple advantages:
- it has good sensitivity and specificity
- it is not affected by prior antibiotic therapy (result is positive for up to a week after the commencement of antibiotic therapy)
- its level is higher in more unwell patients
- it is available earlier than cultures and Gram stains
- Legionella urinary antigen only detects Legionella pneumophila, whereas there are about 19 other species which act as pathogens in man (particularly immunocompromised man). However, it is also positive more often in patients with severe infection. The assay occasionally cross-reacts with Campylobacter antigens.
- FBC: the college recommends we look to the toxic granulation of neutrophils in order to make a diagnosis of infection, remarking that the WCC itself could be low, high, or normal. The BTS expand on this further in their Section 5.8 (iii19), quoting a 1990 paper by Holberg et al. Apparently, among 418 adult Scandinavians, a WCC in excess of 15 was strongly associated with a bacterial (and specifically pneumococcal) pneumonia. It is unclear how this can be generalised to the typical 2015 Western Sydney scenario (150kg bong-smoking diabetic with sleep apnoea) ...but the FBC is probably still important.
- EUC: not only to "modify drugs", as the college suggests. Renal function plays a role in the stratification of mortality risk (the CURB65 score), which in turn recommends for or against ICU admission.
- LFTs are mentioned by the college because "organ involvement, modify drugs, help with aetiology". This is fairly accurate. The BTS guideline mentions LFTs as a means of identifying Legionella cases (which tend to have deranged LFTs). Similarly, leptospirosis presents with a degree of hepatitis.
- Respiratory viral swabs (NAT) - locally, the viral swabs are tested (PCR) for several clinically important viruses, including influenza A and B, parainfluenza, RSV, rhinovirus, adenovirus and possibly one or two others. The BTS suggest that this PCR approach is superior to waiting for serological tests
- Inflammatory markers, eg. CRP and procalcitonin are discussed in the BTS and ISDA guidelines. Salient points made by the societies include the following:
- CRP which fails to fall by 50% byy day 3-4 is associated with an increased 30-day mortality
- CRP levels over 100 were 96% sensitive in identifying pneumonia in one poorly-designed study
- Procalcitonin seems to be a sensitive marker of severe disease and bacterial (especially pneumococcal) aetiology.
Empiric antibiotic therapy for community-acquired pneumonia
2007 ISDA guidelines recommend a broad-spectrum third generation cephalosprin (eg. ceftriaxone 1g bd) and a macrolide (eg. azithromycin 500mg daily). These should be given IV, for a minimum of 5 days (but more likely for 7). Question 2c from the second paper of 2003 obviously predates these guidelines, and recommends old-school erythromycin as the macrolide of choice. A sole agent fluoroquinolone (moxifloxacin) may be substituted.
The most recent NICE guidelines make the following recommendations for the British pneumonia patient with moderate or severe disease:
- 7-10 days of antibiotics
- Dual therapy (a β lactam and a macrolide) for severe pneumonia.
The NICE authors then go on to admit that their recommendations are based more on their experience than on any sort of published evidence, given the quality of what is available.
Causes of a non-response to empiric antibiotic therapy
What is non-response, or 'treatment failure"? I may borrow liberally from the recent 2015 Torres paper, where a series of descriptive criteria were used to determine whether the steroids were working. These criteria were worsening radiological appearance, failure to wean from ventilation (or requiring intubation, or re-intubation), worsening respiratory variables (eg. PaO2/FiO2 ratio), or development of shock. More broady, non-response may be simply the failure of the patient to progress quite as quickly as you would prefer. Question 7 from the second paper of 2012 asks the candidates to "outline the factors that may affect the expected rate of resolution" for a patient with community-acquired pneumonia who has failed to improve after 5 days of mechanical ventilation.
How long do you wait for it to resolve? Even eminent authors admit that the time-based definitions are arbitrary. And when do you start the timer; is it from the first productive cough? From intubation? What if the patient has been having courses of ridiculously inappropriate antibiotics, and has not yet had any treatment per se (so how can he be "treatment refractory"?).
A sufficiently severe pneumonia may be slow to resolve for a variety of reasons which are not at all associated with inappropriate management. A recent article (Meiling et al, 2014) has found that while poor antibiotic choice plays a role, multilobar infiltration and a high CURB-65 score were important independent risk factors for a treatment-refractory slow-to-resolve pneumonia. Not surprisingly, it has a substantially increased mortality associated with it (according to LITFL, five-fold). In total there are several factors which are known as predictors of a poor response to antibiotics; these are listed in the excellent 2007 article by Menendez and Torres.
The table offered below is compiled from college answers to the previous SAQs on this topic, as well as from the abovequoted papers.
Wrong antimicrobial agents
Predictors of poor response to antibiotics:
Investigations for the non-responding pneumonia
A brilliant article on this topic is offered from Clinics in Chest Medicine (Kuru and Lynch, 1999), but unfortunately it is behind a paywall. The next best source is probably the UpToDate page on nonresolving pneumonia. Again, access to the latter requires the exchange of money. This LITFL article, however, is free.
- Culture again! You have selected some sort of Horrendomonas with your empirical therapy, and it will require a different antibiotic cocktail.
- TTE: the contribution of cardiogenic pulmonary oedema to the respiratory failure needs to be considered.
- CT chest; particularly high-resolution CT: it will reveal the fll extent of the pneumonia, and it will unveil new cavitating lesions, loculated collections and bronchial masses.
- Sputum eosinophils: eosinophilic pneumonitis may be to blame.
- Acid-fast bacilli: it would be embarrassing to miss tuberculosis
- Aspergillus investigations as well as the other fungi
- CMV, VZV, HSV - PCR on sputum (though inlikely in an immunocompetent host)
- Autoimmune screen; perhaps this "pneumonia" is in fact a pulmonary manifestation of a systemic autoimmune disease, eg. SLE, RA, Sjögren's syndrome, mixed connective tissue disease, Wegener's granulomatosis, Churg-Strauss syndrome, Goodpasture's syndrome, ankylosing spondylitis, and so on and so forth.
- Bronchoscopy: it will reveal any bronchial obstruction, and it may allow the lavage of a lobe, thereby collecting valuable specimens.
- Lung biopsy: Even though this is invasive, it may be indicated in situations where the diagnosis is uncertain and the potential treatments are aggressive and mutually incompatible (eg. high dose steroids vs. high dose antibiotics)
De-escalation of antibiotic therapy for pneumonia
The 2007 ISDE guidelines (used to answer Question 18 from the first paper of 2012) are:
- Minimum of 5 days
- Stop after 72 hrs of haemodynamic stability without fevers
- Pseudomonas typically needs 15 days
The newer 2014 NICE guidelines (for ICU-level severe pneumonia) recommend 7-10 days of antibiotics; however nothing else is said about this subject. The procalcitonin RCT mentioned in the SAQ is a famous study where antibiotic therapy was deescalated according to a procalcitonin level (anything less than 0.25mic/L was grounds for deescalation).
Corticosteroids for pneumonia
Though the 2009 BTS guidelines did not favour the use of steroids, they did admit that their evidence for this opinion was of particularly poor quality, and that the whole thing resembled gut instinct rather than a well-reasoned recommendation. Fortunately, a flurry of interest in using steroids for pneumonia was aroused after two new trials were published in 2015. Torres et al investigated high-acuity ICU patients, whereas Blum et al looked at the broader population of hospital in-patients. In the severely ill group, steroids decreased the chances of "treatment failure" as defined by deteriorating circulatory or respiratory function, but without much of an impact on mortality. The publication of these trials stimulated the college to ask about this issue in Question 5 from the first paper of 2016.
Rationale for the use of steroids
- Much of the organ system effects of pneumonia may be related to the inflammatory reaction
- Anti-inflammatory drugs like steroids may reduce the release of cytokines, thereby dampening the SIRS response
- Modulation of the pro-inflammatory response should lead to a more rapid resolution of clinical features
- Similar rationale has been successfully applied to other infectious illnesses, for instance in the use of dexamethasone for pneumococcal meningitis
- Other pulmonary infections benefit from corticosteroids, most notably PJP.
- Patients with pneumonia may have underlying COPD or asthma which may be unrecognised in the community, and which will improve with steroids as a "collateral benefit".
Conventional applications of steroids for specific respiratory scenarios
- COPD complicated by pneumonia
- Pneumocystis jirovecii pneumonia (PJP): even in non-HIV patients, around 30mg/d of prednisone seems to reduce ICU stay and duration of mechanical ventilation (Pareja et al, 1998).
- Adrenally suppressed patients (eg. chronic steroids)
- Relative adrenal insufficiency in septic shock: this is discussed elsewhere
- Cryptogenic organising pneumonia (1mg/kg/day of prednisolone, according to a multinational guideline from 2008)
- Fibro-proliferative phase of ARDS - but the evidence is not convincing; there may be some sort of effect if the steroids are given between 7 and 1 days following onset, accroding to Steinberg et al (2006)
- Acute eosinophilic pneumonia ( Davis et al, 1986)
Arguments against the use of steroids
- The systemic inflammatory response to pneumonia has constructive facets, as it mobilises the immune retaliation against the pathogens
- The metabolic side effects (eg. hyperglycaemia, hypernatremia, adrenal suppression) may be a disadvantage to severely septic patients
- The possibility of exacerbating muscle weakness may be a disadvantage with using corticosteroids in mechanically ventilated patients
- Some forms of pneumonia are known to get worse with steroids, eg. Influenza where mortality is doubled (Rodrigo et al, 2015 ) and Aspergillus (Parody et al, 2009 - although this was a study of bone marrow transplant patients)
Evidence regarding use of steroids
- Torres et al : compared methylprednisolone (n = 61) or placebo (n = 59) for 5 days.
- Lless treatment failure with steroids (13% vs 31%)
- No difference in mortality
- Blum et al: 785 patients randomised to 50mg prednisolone or placebo
- "Median time to clinical stability was shorter in the prednisone group" - 3.0 days vs. 4.4
- Not specifically ICU patients - all CAP admissions were enrolled
- A 2015 meta-analysis including the above data suggested that the use of corticosteroids may reduce
- mortality by 3%
- need for mechanical ventilation by 5%
- hospital stay by 1 day
Reception by the ICU community
"Time to change practice", writes Djillali Annane, the great champion of using steroids for everything. "Not for everyone", writes Richard Wunderink, citing some valid concerns regarding the methodology of this recent 2015 Torres study. For instance:
- The placebo group has more intubated patients, and more septic shock
- A surprisingly small number (~ 20-25%) of the patients received proper dual antibiotic therapy
- Only 57% of the patients had a CRP above 15
- The recruitment was weird: the study went on for about 8 years, recruiting an average of 5 patients per center per year.
- The primary treatment difference between groups was radiographic progression between days 3 and 5, which may represent something completely clinically irrelevant.
Reception by the CICM examiners
The college answer to Question 5 from the first paper of 2016 had mentioned that Up-to-Date had included this in one of their practice-changing updates (Nov. 2015). UpToDate now recommend steroids for CAP with a Grade 1B recommendation. However, "not yet accepted therapy" and "treatment effect small" are the summary statements made by the college in their model answer.
Mechanical ventilation for pneumonia
Folk wisdom in the ICU has historically been to view NIV as a prelude to intubation, as far as pneumonia is concerned. The reasoning behind this opinion is the impediment to normal cough which results from the use of NIV, particularly in the frail elderly. If you cannot clear your purulent sputum, you cannot achieve source control, and therefore all the effort-of-breathing improvements achieved with NIV end up being for nought. There have never been questions regarding this specific issue in the CICM Part II, and in any case the answer would belong somewhere in the NIV chapter of the mechanical ventilation section. In brief, all evidence suggests that the only pneumonia patients who benefit from NIV are those who also happen to have a coexisting NIV-treatable disease (eg. acute pulmonary oedema or COPD).