By Jake Scott, MD
Clinical Assistant Professor, Infectious Diseases and Geographic Medicine, Stanford University School of Medicine; Antimicrobial Stewardship Program Medical Director, Stanford Health Care Tri-Valley
SYNOPSIS: Staphylococcus aureus bacteremia (SAB) is associated with high mortality and often is complicated by metastatic sites of infection that can be difficult to identify and control. Fluorine-18 fluoro-2-deoxy-D-glucose positron emission tomography/computed tomography (18F-FDG-PET/CT) has shown promise in some observational studies as an important diagnostic modality for detecting foci of infection and potentially guiding clinical management in ways that may lead to improved outcomes, including lower mortality. The validity of these findings has been questioned, however, due to the risk of significant bias. Researchers conducted a study on the effects of 18F-FDG-PET/CT on mortality in a cohort of patients with SAB while controlling for confounding and immortal time bias.
SOURCE: van der Vaart TW, Prins JM, van Werkhoven CH, et al. Positive impact of [18F]FDG-PET/CT on mortality in patients with Staphylococcus aureus bacteremia explained by immortal time bias. Clin Infect Dis 2023;Mar 4: ciad 112. doi: 10.1093/cid/ciad112. [Online ahead of print].
Van der Vaart and colleagues analyzed existing data from a prospective cohort study previously conducted in seven hospitals in the Netherlands that had recruited adult patients with one or more blood cultures positive for Staphylococcus aureus to examine the effect of 18F-FDG-PET/CT on 90-day mortality. The primary outcome was 90-day all-cause mortality, and the secondary outcome was 90-day infection-related mortality. Patients were excluded from the analysis if they had died within 48 hours of collection of the first positive blood cultures, since 18F-FDG-PET/CT was considered to be highly unlikely to have been performed during that brief interval. 18F-FDG-PET/CT was available at all participating hospitals and was conducted at the discretion of the treating physician. Patients were followed for 90 days after the first day of positive blood cultures through telephone interview, contact with primary care physicians, and review of electronic health records. To adjust for immortal time bias, 18F-FDG-PET/CT was modeled as a time-varying covariate by including all patients in the no-18F-FDG-PET/CT group until they underwent 18F-FDG-PET/CT. Other variables associated with all-cause mortality were adjusted for as confounders and modeled as time-fixed covariates. To examine the effect of not adjusting for immortal time bias, the study also performed a model where 18F-FDG-PET/CT was not a time-varying covariate.
Out of 476 patients determined to be eligible for the study, 178 (37.4%) underwent 18F-FDG-PET/CT. The median time from positive blood cultures to undergoing evaluation by 18F-FDG-PET/CT was nine days (interquartile range [IQR], six to 13 days). Those who underwent 18F-FDG-PET/CT were more likely to have had community-acquired bacteremia (48% vs. 26%), any implanted prosthetic material (46% vs. 35%), high-risk bacteremia (67% vs. 39%), diagnosis of endocarditis (29% vs. 12%) or osteoarticular infection (29% vs. 15%), longer duration of antimicrobial therapy (median 42 vs. 16 days), and source control intervention performed (44% vs. 34%). Based on a crude analysis without controlling for variables associated with all-cause mortality, 18F-FDG-PET/CT was associated with a reduction in all-cause mortality (univariate hazard ratio [HR], 0.59; 95% confidence interval [CI], 0.41 to 0.86). This association remained relatively unchanged after adjusting for known predictors of all-cause mortality (adjusted hazard ratio [aHR], 0.50; 95% CI, 0.34 to 0.74). This result is comparable to results from other observational studies that had adjusted for particular confounders but not for immortal time bias.
After adjusting for immortal time bias, by modeling 18F-FDG-PET/CT as a time-dependent variable, van der Vaart and colleagues found no protective association between 18F-FDG-PET/CT and all-cause mortality (HR, 1.07; 95% CI, 0.72 to 1.57). Further adjusting for additional confounding variables (age, comorbidities, positive surveillance blood cultures at 48 hours, septic shock at time of presentation, and presence of endocarditis) also failed to demonstrate a protective association between 18F-FDG-PET/CT and all-cause mortality (aHR, 1.00; 95% CI, 0.68 to 1.48), nor any association between 18F-FDG-PET/CT and infection-related mortality (aHR, 1.30; 95% CI, 0.77 to 2.21).
COMMENTARY
While 18F-FDG-PET/CT may serve as an important tool in particular instances for detecting metastatic foci in patients with SAB, the extent to which its use leads to significant changes in clinical outcomes remains uncertain.1,2 A recent review identified four observational studies that reported an association between receipt of 18F-FDG-PET/CT and reduced mortality ranging from 13.1% to 27.8% in patients with SAB or other gram-positive bacteremia.3 All observational studies are susceptible to some degree of confounding, which varies based on the statistical methods applied. Immortal time bias, also referred to as time-dependent bias, occurs when an analysis does not account for a time-dependent intervention, such as the receipt of a diagnostic procedure, that requires that the patient survive long enough to receive the intervention.4 Immortal time bias often is overlooked in observational studies, but the magnitude of effect that immortal time bias may have on results can be quite significant, as this study by van der Vaart and colleagues clearly illustrates. Since mortality rates attributed to SAB are highest early in the disease course and decrease with time, correcting for immortal time bias in these types of studies is essential.
REFERENCES
- Berrevoets MAH, Kouijzer IJE, Aarntzen EHJG, et al. 18F-FDG PET/CT optimizes treatment in Staphylococcus aureus bacteremia and is associated with reduced mortality. J Nucl Med 2017;58:1504-1510.
- Yildiz H, Reychler G, Rodriguez-Villalobos H, et al. Mortality in patients with high risk Staphylococcus aureus bacteremia undergoing or not PET-CT: A single center experience. J Infect Chemother 2019;25:880-885.
- Thottacherry E, Cortés-Penfield NW. Evidence of clinical impact supports a new petition for Medicare coverage of 2-[18F]fluoro-2-deoxy-D-glucose positron emission tomography/computed tomography in the evaluation of Staphylococcus aureus bacteremia: A focused literature review and call to action. Clin Infect Dis 2022;75:1457-1461.
- Jones M, Fowler R. Immortal time bias in observational studies of time-to-event outcomes. J Crit Care 2016;36:195-199.
