Of coccus, Rocco and scores: pneumococcal disease, Rocky Graziano and pneumonia severity scoring systems

View addendum:

Non-bacteremic pneumococcal pneumonia: general characteristics and early predictive factors for poor outcome

The reader of Infectious Diseases can find in this issue an interesting analysis by Lyere Serrano Fernandez et al. [1] on general characteristics and early predictive factors of poor outcome in non-bacteremic pneumococcal pneumonia.

Community-acquired pneumonia (CAP) is the leading cause of death from infectious diseases and remains an important public health problem, with significant implications for healthcare systems worldwide. Despite important advances in prevention via more effective and widespread use of vaccines, novel rapid tests for diagnosis, and early use of antibiotics, management of CAP continues to pose significant challenges. Despite wide geographic variations, mortality remains high, especially in the severely ill presenting with respiratory failure or shock, in the elderly or in those with significant medical comorbidities. Even after an acute episode of CAP, higher risk of cardiovascular events or death remain for a significant period of time, a risk likely driven by persistent or rebound in inflammation, in active interaction with the host’s coexistent conditions and risk factors.

In usual practice, 40–60% of CAP cases have identifiable aetiologies. Advanced laboratory techniques and rigorous assessments can identify an etiologic agent in up to 40–60% of the remaining cases, which roughly follow the same distribution of causative microbes (bacteria and viruses mostly). Streptococcus pneumoniae (pneumococcus, gram positive, lancet-shaped coccus or diplococcus) is the leading cause of pneumonia worldwide, being responsible for the disease in 40–60% of causally identified cases (familiar theme? easy to remember, for sure!). There have been secular trends of pneumococcal disease decline, especially in the western world, possibly due to the use of vaccines and to our improved ability to establish an etiologic diagnosis, i.e. identifying other causative agents.

Pneumococcal infections are often subclassified into non-invasive and invasive disease. Non-invasive pneumococcal disease is represented by sinusitis, acute otitis media and pneumonia. Invasive pneumococcal disease is defined as an infection confirmed by the isolation of the microbes from a normally sterile site, such as blood or cerebrospinal fluid, and it consists mainly of pneumococcal meningitis, bacteremic pneumonia and pneumococcal bacteraemia without a primary focus (also called in the past isolated ‘benign’ pneumococcal bacteraemia, although nothing benign about it). The clinical spectrum of pneumococcal CAP varies widely, from a near-asymptomatic phenotype in some cases, to a devastating illness, complicated by respiratory insufficiency, septic shock and occasionally multi-organ failure in others. Septic shock remains one of the most important prognostic determinants in patients with pneumococcal CAP. In 40–60% of the CAP cases, patients could not expectorate sputum for microbiologic examination; among the 40–60% of those who do, 40–60% of the specimens are technically inadequate (sic!) [2,3]. Generally, blood cultures are positive in less than 1 in 6 hospitalized patients with CAP, while the frequency of a positive test increases with the disease severity [4–6].

Only a couple of decades ago, sputum and blood cultures represented the main routine diagnostic tests for CAP. Interestingly, several studies found that these conventional diagnostic tests did not contribute significantly to the overall patient disease management [7–9]. More recently, urinary antigen tests to diagnose pneumonia caused by Legionella pneumophila (serotype 1) or Streptococcus pneumoniae have been introduced. An immunochromatographic assay has been developed to detect C polysaccharide cell wall antigens that are present in all pneumococci, a very useful rapid test for the diagnosis of CAP, with sensitivities and positive predictive values in the 50–80% range, while specificity and negative predictive values have been in the 90%s. The test sensitivity is higher in patients with bacteremic pneumonia, and the test does not appear to be influenced by prior use of antibiotics. The performance of this diagnostic test is occasionally affected by other, non-pneumococcal species (false positive tests), recent pneumococcal disease (with persistently positive results) or nasopharyngeal carrier status (especially in children). It is important to know that no organism is effectively isolated, hence no antibiotic sensitivities are obtained in these cases [10,11].

With the intent of using at the point-of-care models of unfavourable outcomes or to predict those at higher risk of dying from the disease, numerous pneumonia severity scoring systems have been developed and published before, from PSI [12], ATS [13], modified ATS [14], BTS (CURB) [15], CURB-Age [16], modified BTS (CURB‐65) [17], CRB-65 [18], CORB [19], SMART‐COP [20] etc. Some of them have been tested and validated for all CAP, others specifically in pneumococcal pneumonia. The good news is that mortality in both bacteremic and non-bacteremic pneumococcal CAP seem to be declining, to less than 17% and 8%, respectively. Nevertheless, the challenge to find the ideal stratifying tool for severity or mortality in both groups may be elusive and may not replace the value of individualized ‘art’ or that of smart doctors…

The report by Serrano Fernandez et al. [1] is likely the largest (n = 638) prospective study of patients with non-bacteremic pneumococcal CAP confirmed by urinary antigen assays, which aims at studying predictors of outcome and risk stratification. In order to avoid ‘floor effects’ due to low mortality rates in non-bacteremic pneumonia, Serrano Fernandez et al. [1] propose to use an aggregate primary outcome measurement called ‘poor outcome’, which includes inpatient mortality, need of mechanical ventilation or shock. Not surprisingly, patients with observed poor outcomes were more often nursing home residents, individuals with significant habitual ethanol intake, smokers and those who presented with tachypnoea, hypotension, tachycardia, altered mental status, more laboratory and radiological abnormal findings, or overall worse disease severity upon admission. The poor outcome was found in approximately 13% of the cohort. The authors performed both univariate and multivariate logistic regression looking for predictors of poor outcome and underwent an internal validation of their model with simple nonparametric bootstrapping using 1,000 bootstrap samples, which confirmed the robust model, with an AUROC of 0.867 and very small errors.

Most of the predictive variables of poor outcome in the model built by Serrano Fernandez et al. [1] are already included in the Pneumonia Severity Index (PSI) [12] or Confusion Urea Respiratory rate and Blood pressure (CURB) [15] scoring systems. However, approximately 6% of patients in low PSI class on admission had severe clinical course, roughly a quarter of the total number of their patients with poor outcome. The authors correctly suggest that this may reflect the fact that PSI is not sensitive enough to predict poor outcomes in lower risk groups. This limitation has been reported before and is in fact a reason why the use of PSI as an emergency room decision tool for hospitalizing the patient may be imperfect. Two variables not included in the PSI – leukopoenia (peripheral blood leukocytes <4,000 cells/µL) and severe inflammation, the latter defined as CRP ≥15 mg/L and blood leukocytes ≥30,000 cells/µL, represented additional predictors of poor outcomes.

The study by Serrano Fernandez et al. [1] confirms a low inpatient mortality (only 4.1%) in those with non-bacteremic pneumococcal CAP, lower than that of bacteremic pneumonia, the latter having an odds ratio of dying in the hospital of 2.1 (95% CI 1.1–3.9, p = .02), as previously published by some of the same group [21]. Similarly, others have found a risk ratio of dying in the hospital of 1.63 (95% CI 1.06–2.50, p = .026) for bacteremic vs. non-bacteremic subjects [22]. Interestingly, only 22.7% and 32.3% of this study’s participants had been previously vaccinated with pneumococcal or influenza vaccines, respectively. Among the vaccinated individuals, pneumococcal vaccine was mostly of 23-valent type (as available in Spain during the study enrolment years), which is now known to not influence as much the rates of incident non-bacteremic pneumonoccal CAP [23,24]. The authors propose that early identification of patients with likely poor outcome can lead to care optimization and possibly to interventions that could avert an adverse outcome in hospitalized immunocompetent patients with non-bacteremic pneumococcal pneumonia. If confirmed in an external cohort, this can help identify poor outcomes in the more restricted pneumococcal CAP category diagnosed with positive urinary antigen tests and negative blood cultures. While the study presents a large cohort of consecutive patients from two different hospitals in the same region of Spain, future observational studies should expand the geographic footprint and take an approach of a priori study planning based on propensity scoring or clustering, as treatments and outcomes are more intricately intertwined. For example, one patient with higher comorbid load and initial high severity may receive antibiotics faster, intravenous fluids more aggressively, may be treated with a combination of intravenous antibiotics, receive more attentive care, and still have worse outcomes. Simplifying, this does not lead to the apparent conclusion that a combination of intravenous antibiotics is responsible for worse outcomes, but effectively that more severe patients (sometimes even based on gestalt clinical impression) are treated differently, clinicians may have lower threshold of admitting or transferring to medical intensive care unit, instituting aggressive early sepsis management, etc.

Thomas Rocco Barbella (1919–1990), better known as Rocky Graziano, was an American professional boxer who once held the World Middleweight champion title [25]. Rocco is still considered one of the most brutal knockout (KO) ‘artists’ in boxing history, often displaying the ability to take out his opponent with a single punch. His professional boxing score record was 67 wins (52 by KO), 10 losses and 6 draws. In comparison, even if one ignores for a second the horrible devastation created by the recent novel coronavirus pandemic, pneumonia remains a very serious condition and has the potential to lead to the host’s KO in one ‘punch’. Rocco is credited to have once said: ’I quit school in the sixth grade because of pneumonia. Not because I had it, but because I couldn’t spell it’. Similarly, being able to spell out the CAP risk factors and outcome predictors from various scores is a very important endeavour for all of us, so that we become more effective in formulating and implementing tailor-made, personalized therapeutic approaches for our patients with pneumococcal pneumonia.

Disclosure statement

No potential conflict of interest was reported by the author(s).