Community-acquired pneumonia: The need to broaden our diagnostic armamentarium

Community-acquired pneumonia (CAP) is associated with significant mortality and morbidity, and high costs globally. The epidemiology of CAP demonstrates high regional and geographic variability with regard to age, risk factors, disease severity and causative pathogens.

to the emerging global picture, this South African (SA) study confirmed that viral pathogens were more common than bacterial pathogens, with rhinovirus being most commonly identified. [3,[5][6][7] In this issue of the AJTCCM, researchers from Tygerberg Hospital (TBH), one of two large tertiary hospitals in the Western Cape Province, reviewed their admissions for CAP to the intensive care unit (ICU) over a 12-month period. [8] They found that patients with CAP accounted for 17.5% of the 423 admissions during that period, confirming the high burden of CAP in this setting.
However, although requiring ICU admission, the APACHE II score ranged from 6 to 39, with an overall ICU mortality rate of 21.6% -lower than the reported CAP mortality in ICUs in other settings. [1][2][3]9] As with many studies of mortality in ICUs, this may very well represent local resource limitations, and varying thresholds for ICU admission. Nonetheless, the results are reassuring for a setting where the average age of patients with CAP is several decades lower than in most of the developed world, and in which pneumonia is a leading cause of death.
In keeping with published data from other parts of the country, TB was found to be a relatively common pathogen, accounting for roughly 22% of CAP admissions, representing the most common identifiable cause. This confirms the increasing presentation of TB as an acute pneumonia. [10] Mazaza et al. [8] argue, and rightly so, that we should actively confirm or exclude TB in every patient admitted to the ICU. The location of their study, in the global epicentre of TB, reminds us that we should always seek to 'know our epidemic, and know our response' -the global rallying call for intensified local responses to global epidemics based on unique local disease epidemiology. Failing to identify TB as a cause of pneumonia in the ICU, conversely, may have disastrous consequences and may contribute to the ongoing high mortality of this curable infectious disease. The risks to ICU staff pursuant to TB transmission dynamics may be mitigated by the fact that most patients are managed on closed ventilation circuits. However, the implications for infection control remain significant. Adequate safety precautions, including, but not limited to, isolation of TB cases, must be available and carried out to prevent nosocomial transmission. And most importantly, an improved understanding of TB drug pharmacokinetics and optimised dosing in critical illness is desperately needed.
What is evident from this study is that the vast majority of cases of CAP admitted to an ICU had no pathogen identified. However, molecular tests for viral pathogens M. pneumoniae and C. pneumoniae, and urine testing for S. pneumoniae and Legionella spp., were not routinely performed. This may have significantly compromised the claim of a systematic evaluation for a causative pathogen in this cohort. [5,7,11,12] In addition, the timing of sampling in relation to antibiotic exposure is not explicit, and is likely to have reduced the diagnostic yield of the performed tests. This is especially likely given the increasing pressure on emergency units to initiate antibiotics as soon as sepsis is suspected, which is usually prior to ICU admission.
Another interesting finding was the relatively common identification of highly susceptible P. aeruginosa in 12% of patients. No clear explanation for this occurrence is provided, but it is likely that P. aeruginosa is more Community-acquired pneumonia: The need to broaden our diagnostic armamentarium ubiquitous in the community than we had thought. This low prevalence of phenotypic resistance was also recently reported by the sentinel surveillance laboratories of SA, and may help inform future empirical therapeutic decisions in patients with suspected P. aeruginosa infection. [13] What is clear from this study is that if you do not test for a pathogen, you will not identify it. The ICU at TBH routinely tests all patients for TB using the GeneXpert test. The GeneXpert MTB/RIF assay, however, is not infallible, with an imperfect sensitivity and a clinically significant rate of false-positive results, especially in patients with previous TB. The authors do not present culture confirmation of the GeneXpert resultsnor what percentage of GeneXpert-positive patients had prior TB, in order to confidently exclude false positive tests.
The findings of this study from a large SA hospital contribute to the arduous search for pathogens in our local burden of CAP. This information may inform empirical treatment choices, but needs to be expanded on by methodologically rigorous studies in this field. The challenge of developing a locally responsive empirical strategy is in balancing the risk of failure to treat, on the one hand, and the risks of overtreatment on the other. [9] What is made clear by Mazaza et al. [8] is that unless we start looking, we may never know what we are up against, or how to tailor our response.