Blood culture time-to-positivity can identify S aureus, but not Enterococcal or Streptococcal endocarditis.

Background

Staphylococcus aureus bacteremia (SAB) is a common and important cause of infective endocarditis (IE). Currently, the IDSA guidelines for SAB recommend echocardiography for all patients; however, it has been proposed that time-to-positivity (TTP) could be used to predict and stratify patients more likely to have endovascular infections such as IE^1,2. One publication of 1703 patient episodes showed that IE reduced by 11% per hour increase in TTP and gave a negative predictive value of 96% at times beyond 13 hours³. Indeed, TTP, in addition to the clinical decision tool VIRSTA, improves the negative predictive value to 100%⁴. This suggests that obtaining echocardiography for all patients with SAB with a blood culture time to positivity over 13 hours without clinical evidence or specific risk factors for endocarditis may not be necessary.

Other organisms associated with IE include E faecalis (13.6%)⁵ and some species of high-risk streptococci⁶, but no clear clinically useful association has been reported between IE and TTP in these organisms^7,8, with one study giving an ROC AUC of 0.75 for TTP on IE diagnosis and an association with mortality rates⁹. TTP is dependent on the generation time of the organism, which is comparable across these causes^10–12; and inoculum size, which should be consistent across the same disease process. It is reasonable to explore whether TTP can discriminate between those with and without. These organisms are common causes of bacteremia; however, currently, there is no recommendation to perform echocardiography in all patients, and the European Society of Cardiology 2023 Endocarditis guidance highlights the potential utility of clinical decision tools for the organisms¹³. With IE prevalence as high as S. aureus, the decision to pursue IE investigations is based on clinical suspicion.

To investigate whether these results are generalizable to cohorts in the UK and to organisms other than S. aureus, we retrospectively investigated our cohort of patients with bacteremia caused by S. aureus, high-risk streptococci and enterococci.

Methods

Statistical analysis

Descriptive data of organisms were reported, including positivity rate, mean and median time-to-positivity, and number of cases. Bacteremia was excluded if the time to positivity data was missing or species identification was unavailable. Mann-Whitney tests were performed on each species for IE- and non-IE-associated bacteremia. We explored the association between TTP and IE risk in a) a logistic regression model, calculating the increased risk for each hour of growth, b) a logistic regression model, binarizing TTP at the median TTP, and c) fitting a non-linear model (GAM) to explore inflection points and potential non-linearity. The predictive value of the median TTP was calculated. As transport time to the laboratory can influence time to positivity, we used our cohorts’ median TTP rather than median values reported in the literature¹⁵.

Results

Eight hundred and fifty-six bacteraemias were included, of which 46 were removed due to missing TTP data. There were 159 cases of IE: 37 staphylococci (16.2% of 228 bacteremias), 20 enterococci (18.3% of 109), and 97 streptococci (20.5% of 473). The mean TTP was: staphylococci: 14.1 hours (median 13, IQR 10-16); enterococci 11.6 hours (median 11, IQR 9-14); streptococci 14.5 hours (median 13, IQR 10-17). The streptococcal group included cristatus 3, gallolyticus 48, gordonii 13, mitis/oralis 75, mutans 7, sanguinis 34.

The univariate logistic models (IE ~ TTP) produced Odds Ratios of: staphylococci 0.92 (95%CI 0.85-0.99, p:0.02) enterococci 1.08 (95%CI 0.99-1.18, p:0.1) and streptococci 1.02 (95%CI 0.98-1.06, p:0.31) for every increase in hour of TTP. Using growth below the median TTP as a binomial (cutoff) variable produced the OR: staphylococci 2.62 (95% CI 1.25-5.46, p = 0.01), enterococci 0.94 (95% CI 0.36-2.44, p:0.90) and streptococci 0.80 (95% CI 0.52-1.24, p:0.32) (Figure 1).

Figure 1. Relative density plots for each organism, grouped by IE status against the time to positivity in hours.

The red dashed lines indicate the species-specific median TTP.

The negative predictive values for growth beyond the median TTP for IE is staphylococci 90% (95% CI 84-95), enterococci 80% (95%CI 68-90) and streptococci 78% (95% CI 73-83). The Generalized additive models exhibited monotonicity and no nonlinearity. These results are shown in the supplementary material.

Discussion

Our main findings demonstrate that S. aurues bacteremia with longer TTP is significantly less likely to be associated with IE (8% per hour), but this was not true for enterococci or streptococci. Here, S. aureus taking > 13 h to grow had a Negative Predictive Value of 90% (95% CI 85-95). Our study matches the values previously reported by Stromdahl et al., despite the inability to analyze confounding factors in our cohort. This finding highlights the strength of this association. It will be reassuring to infection specialists that patients who cannot tolerate transesophageal echocardiography or are unable to obtain timely imaging can still be risk assessed for endocarditis prior to completing the 2023 ISCVID Duke scoring.

The lack of a relationship between time-to-positivity for streptococci and enterococci may reflect the differences in IE clinical syndromes with these organisms, typically a subacute presentation as opposed to an acute presentation with S. aureus. This assumes that TTP is related to the bacterial density of the original inoculum, which could be due to lower circulating CFU or sample volume¹⁶. Notably, the risk of IE remains above 15% across all TTP values for these groups, and there is no clear relationship between the two. While the species-specific analysis of streptococci suggests that S. sanguinis and S. gordonii might have a statistical difference in TTP per IE status, there is a clear overlap in the groups, and clinical significance is not met (Figure 2).

Figure 2. Violin plots of time to positivity for each Streptococcus species, with mean time to positivity for IE status compared by Wilcoxon ranked sum testing and p values displayed above.

S cristatus removed due to low numbers of positive cases.

We were not able to control for antibiotic exposure prior to sampling, and we were only able to duplicate the culture results from within the hospital network. IE was diagnosed by an infection specialist in all cases, but 2023 ISCVID Duke's criteria were not extracted in this study.

Conclusion

A longer TTP is associated with lower rates of IE in S. aureus bacteremia at 8% reduction per hour, giving an NPV of 90% beyond 13 hours. There was no evidence of an association with enterococci or streptococci.

Ethics

Four reviews of bacteremias were undertaken as part of quality improvement projects and granted local permission to collect patient identifiable data for these purposes (North Bristol Trust, and University Hospitals Bristol and Weston). Data taken from these reviews included only the TTP results and IE diagnosis. Therefore, formal ethics approval was not required.