A new study published on the preprint server medRxiv* in October 2020 shows that the baseline levels of inflammatory markers and C-reactive protein (CRP) in relation to antibiotic administration can rule out bacterial co-infections and help reduce the unnecessary use of antibiotics.
As the COVID-19 pandemic continues to challenge public health worldwide, it has become clear that a dysregulated and hyperactive inflammatory response is linked to severe and critical disease and a higher mortality rate. Earlier research has shown that flu and other respiratory viral illnesses can be complicated by bacterial co-infections, which can worsen inflammation and result in a higher mortality rate.
At present, few researchers think this to be the case with COVID-19 since most patients have failed to show evidence of bacteria in respiratory samples. The most commonly used tool is a microbiological culture, but this takes several days to produce a result. It is not sensitive enough and may produce confusing results by not distinguishing bacterial colonization from true infection. It is not carried out on respiratory tract samples on a routine basis in COVID-19 patients.
This has led to a high incidence of unwarranted antibiotic use in these patients, which could result in antibiotic-resistant strains of bacteria posing an increasing threat in the days ahead.
Study: Exclusion of bacterial co-infection in COVID-19 using baseline inflammatory markers and their response to antibiotics. Image Credit: Kateryna Kon / Shutterstock
Using Novel Markers to Differentiate Bacterial vs. Viral Infection
A lot of research has gone into attempts to identify clinical and laboratory features that can classify patients with COVID-19 on the basis of the predicted outcomes. However, it is hard to differentiate between subjects with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection alone and those with a superimposed bacterial infection.
In influenza, earlier studies have explored the use of the inflammatory markers C-reactive protein (CRP), white cell count (WCC), and procalcitonin (PCT) to pick up bacterial co-infections. Measurements of these markers at specified intervals may also help in this task. While some success has been shown using neutrophil counts at admission, along with D-dimer and CRP levels, a clear definition of the changes that are predictive of bacterial pneumonia is lacking.
The current study explores this area to find clear discriminants for viral COVID-19 infections vs. those with bacterial co-infection. The parameters for bacterial pneumonia were defined using the results obtained in laboratory tests in patients with community-acquired pneumonia (CAP). These were then compared with the baseline inflammatory markers and their response to antibiotics to determine if this could provide a specific differentiation between the two.
The data came from a retrospective cohort study in a large UK teaching hospital that is located on two different campuses. The researchers found that of the patients with COVID-19, 62% were male but only about half in the cohorts with influenza or CAP. Again, Black, Asian, Mixed, and Other (non-white) ethnicity patients were disproportionately affected by COVID-19. The median age of CAP and COVID-19 patients was higher compared to those with influenza. Comorbidities were higher in CAP patients, who also were more likely to have positive cultures for bacteria.
Distinguishing Pneumonia from COVID-19
The researchers found that on comparison of inflammatory markers, the total white cell count, differential count, and CRP levels were highest in CAP vs. COVID-19 and influenza. The white cell counts were at a median of ~12.5, 6.8, and 7.2 x106/mL, respectively. Neutrophils counts showed the same association, but not lymphocyte counts.
The CRP level was higher in CAP, compared to COVID-19, at median levels of 133.5 and 86 mg/L, respectively. It was lowest in influenza, at 31 mg/L.
Antibiotics were used for the treatment of all patients with a diagnosis of CAP, as expected. However, 95% and 88% of COVID-19 patients were also treated with antibiotics, as found in two separate surveys of patients on these campuses, citing the presumed presence of bacterial co-infection.
Using this data, the researchers looked at the changes in inflammatory markers following the initiation of antibiotic therapy in these cohorts. They found that about half of patients with either CAP or COVID-19 stayed in hospital for more than 48 hours and had at least one blood draw between 48-72 hours later.
At this point, the inflammatory marker levels showed significant associations. Patients with CAP had high median white cell counts, but there was a significant fall from baseline. There was also no longer a significant difference between the CRP in CAP and COVID-19 patients. These changes were due chiefly to a pronounced reduction in white cell counts and CRP levels in CAP patients compared to COVID-19 patients.
While the white cell count dropped by a median of -2.3 x106/ml in CAP patients, the decline in COVID-19 patients was only 0.17 x106/ml, after starting antibiotics. The median CRP levels in CAP and COVID-19 were 107.5 vs. 127.0mg/L, and no longer statistically significant.
Using Multiple Discriminant Variables
These findings suggest that the use of baseline white cell counts and CRP, and a follow-up test at 48-72 hours, could discriminate between COVID-19 and CAP. The researchers then tried to discover if they could answer “Yes” or “No” to the question of whether a given patient had CAP or not. They found that the use of either white cell count or CRP by itself yielded the most accurate answer.
After adjusting for ICU admission and for the presence of proven bacterial co-infection in COVID-19 patients, they found no significant differences in the observed associations of white cell counts or CRP with CAP and COVID-19.
Discriminant Criteria: Bacterial Pneumonia vs COVID-19
The researchers then identified the cut-offs for the values of white cell counts and CRP levels that could best distinguish CAP from COVID-19. They found that using a combination of criteria, namely, both a WCC>8.2x106/ml and ΔCRP<0, could pick up 90% of cases of bacterial infection, but with a low specificity of only 43%.
Yet, they comment, “The absence of both admission WCC>8.2x106/ml and ΔCRP<0 could still exclude CAP, and by extension bacterial co-infection alongside COVID-19, promoting antibiotic cessation in 43% COVID-19 patients from this cohort.”
These findings were validated in three cohorts of patients with CAP, COVID-19, and influenza at another hospital. This showed that CAP was correctly excluded in ~46% of COVID-19 patients without positive bacterial cultures or radiological evidence of pneumonia, who were hospitalized for more than 48 hours.
Notably, almost all these patients received a full 5-day course of antibiotics. If these criteria had been applied, the antibiotics could have been stopped at 48-72 hours, preventing 51 days of antibiotics, and reducing the overall antibiotic prescriptions by a quarter.
Implications and Future Directions
The ability to discriminate these two conditions using large groups, based on readily available investigations, in a validated cohort, and the production of easily interpretable criteria, makes this study a valuable one for routine clinical purposes. At the same time, individual hospitals can modify the criteria to create cut-offs that better fit their own laboratory value distributions for CAP. Moreover, these criteria do not negate the role of clinical judgment in decision making, such as considering the patient’s need for oxygen and negative radiological signs in deciding whether antibiotics are necessary in any given case.
Further research will be necessary to compare the use of these criteria with the eventual outcome of these patients. Also, the underlying assumption of this study is that bacterial pneumonia in COVID-19 patients will have the same disease processes and inflammatory profile as in CAP without COVID-19. This requires empirical verification, though the findings of this study do not suggest its incorrectness.
The researchers suggest that the predictive power of PCT (which was not measured routinely in these cohorts and was therefore not part of the current study) should also be explored, along with D-dimer levels and perhaps even transcriptional markers indicating inflammatory cytokines.
The authors conclude: “Routine clinical parameters, admission WCC and changes in CRP following antibiotic administration, can be translated into a set of diagnostic criteria that can exclude bacterial co-infection in up to half of COVID-19 patients.” This could pave the way for rational antibiotic prescription protocols to assist antibiotic stewardship.
medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.