Background: Blood cultures are commonly ordered for patients with low risk of bacteremia. Liberal blood-culture ordering increases the risk of false-positive results, which can lead to increased length of stay, excess antibiotics, and unnecessary diagnostic procedures. We implemented a blood-culture indication algorithm with data feedback and assessed the impact on ordering volume and percent positivity. Methods: We performed a prospective cohort study from February 2022 to November 2022 using historical controls from February 2020 to January 2022. We introduced the blood-culture algorithm (Fig. 1) in 2 adult surgical intensive care units (ICUs). Clinicians reviewed charts of eligible patients with blood cultures weekly to determine whether the blood-culture algorithm was followed. They provided feedback to the unit medical directors weekly. We defined a blood-culture event as ≥1 blood culture within 24 hours. We excluded patients aged <18 years, absolute neutrophil count <500, and heart and lung transplant recipients at the time of blood-culture review. Results: In total, 7,315 blood-culture events in the preintervention group and 2,506 blood-culture events in the postintervention group met eligibility criteria. The average monthly blood-culture rate decreased from 190 blood cultures per 1,000 patient days to 142 blood cultures per 1,000 patient days (P < .01) after the algorithm was implemented. (Fig. 2) The average monthly blood-culture positivity increased from 11.7% to 14.2% (P = .13). Average monthly days of antibiotic therapy (DOT) was lower in the postintervention period than in the preintervention period (2,200 vs 1,940; P < .01). (Fig. 3) The ICU length of stay did not change before the intervention compared to after the intervention: 10 days (IQR, 5–18) versus 10 days (IQR, 5–17; P = .63). The in-hospital mortality rate was lower during the postintervention period, but the difference was not statistically significant: 9.24% versus 8.34% (P = .17). The all-cause 30-day mortality was significantly lower during the intervention period: 11.9% versus 9.7% (P < .01). The unplanned 30-day readmission percentage was significantly lower during the intervention period (10.6% vs 7.6%; P < .01). Over the 9-month intervention, we reviewed 916 blood-culture events in 452 unique patients. Overall, 74.6% of blood cultures followed the algorithm. The most common reasons overall for ordering blood cultures were severe sepsis or septic shock (37%), isolated fever and/or leukocytosis (19%), and documenting clearance of bacteremia (15%) (Table 1). The most common indications for inappropriate blood cultures were isolated fever and/or leukocytosis (53%). Conclusions: We introduced a blood-culture algorithm with data feedback in 2 surgical ICUs and observed decreases in blood-culture volume without a negative impact on ICU LOS or mortality rate.

Disclosure: None

This retrospective review of 4-year surveillance data revealed a higher central line-associated bloodstream infection (CLABSI) rate in non-Hispanic Black patients and higher catheter-associated urinary tract infection (CAUTI) rates in Asian and non-Hispanic Black patients compared with White patients despite similar catheter utilization between the groups.

Background: Central-line–associated bloodstream infections (CLABSIs) arise from bacteria migrating from the skin along the catheter, by direct inoculation, or from pathogens that form biofilms on the interior surface of the catheter. However, given the oxygen-poor environments that obligate anaerobes require, these organisms are unlikely to survive long enough on the skin or on the catheter after direct inoculation to be the true cause of a CLABSI. Although some anaerobic CLABSIs may meet the definition for a mucosal-barrier-injury, laboratory-confirmed, bloodstream infection (MBI-LCBI), some may be not. We sought to determine the proportion of CLABSIs attributed to obligate anaerobic bacteria, and we sought to determine the pathophysiologic source of these infections. Methods: We performed a retrospective analysis of prospectively collected CLABSI data at 54 hospitals (academic and community) in the southeastern United States from January 2015 to December 2020. We performed chart reviews on a convenient sample for which medical records were available. We calculated the proportion of CLABSIs due to obligate anaerobes, and we have described a subset of anaerobic CLABSI cases. Results: We identified 60 anaerobic CLABSIs of 2,430 CLABSIs (2.5%). Of the 60 anaerobic CLABSIs, 7 were polymicrobial with nonanaerobic bacteria. The most common species we identified were Bacteroides, Clostridium, and Lactobacillus (Table 1). The proportion of anaerobic CLABSIs per year varied from 1.2% to 3.7% (Fig. 1). Of 60 anaerobic CLABSIs, 29 (48%) occurred in the only quaternary-care academic medical center in the database. In contrast, an average of 0.6 (SD, 0.6) anaerobic CLABSIs occurred in the 53 community hospitals over the 6-year study period. Of these 29 anaerobic CLABSIs, 23 (79%) were clinically consistent with secondary bloodstream infections (BSIs) due to gastrointestinal or genitourinary source, but they lacked appropriate documentation to meet NHSN criteria for secondary BSI or MBI-LCBI based on case reviews by infection prevention physicians. The other 6 anaerobic CLABSIs did not have a clear clinical etiology and did not meet MBI-LCBI criteria. In addition, 27 (93%) of 29 anaerobic CLABSIs occurred in patients who were either solid-organ transplant recipients, were stem-cell transplant recipients, or were receiving chemotherapy. Lastly, 27 (93%) of 29 anaerobic CLABSIs were treated with antibiotics. Conclusions: Anaerobic CLABSIs are uncommon events, but CLABSI may disproportionately affect large, academic hospitals caring for a high proportion of medically complex patients. Additional criteria could be added to the MBI-LCBI to better classify anaerobic BSI.

Funding: None

Disclosures: None

Background: Racial and ethnic disparities in healthcare access, medical treatment, and outcomes have been extensively reported. However, the impact of racial and ethnic differences in patient safety, including healthcare-associated infections, has not been well described. Methods: We performed a retrospective review analyzing prospectively collected data on central-line–associated bloodstream infection (CLABSI) and catheter-associated urinary tract infection (CAUTI) rates per 1,000 device days. Data for adult patients admitted to an academic medical center between 2018 and 2021 were stratified by 7 racial and ethnic groups: non-Hispanic White, non-Hispanic Black, Hispanic/Latino, Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, and othe. The “other” group was composed of bi- or multiracial patients, or those for whom no data were reported. We compared the CLABSI and CAUTI rates between the different racial and ethnic groups using Poisson regression. Results: Compared to non-Hispanic White patients, the rate of CLABSI was significantly higher in non-Hispanic Black patients (1.27; 95% CI, 1.02–1.58; P < .03) and those in the “other” race category (1.79; 95% CI, 1.39–2.30; P < .001, respectively), and these trends increased in Hispanic/Latino patients (Table 1). Similarly, Black patients had higher rates of CAUTI (1.42; 95% CI, 1.05–1.92; P < .02), as did Asian patients (2.49; 95% CI, 1.16–5.36; P < .02), and patients in the “other” category (1.52; 95% CI, 1.06–2.18; P < .02) (Table 2). Conclusions: Racial and ethnic minorities may be vulnerable to a higher rate of patient safety events, including CLABSIs and CAUTIs. Additional analyses controlling for potential confounding factors are needed to better understand the relationship between race or ethnicity, clinical management, and healthcare-associated infections. This evaluation is essential to inform mitigation strategies and to provide optimum, equitable care for all.

Funding: None

Disclosures: None

Background: Despite significant morbidity and mortality, estimates of the burden of healthcare-associated viral respiratory infections (HA-VRI) for noninfluenza infections are limited. Of the studies assessing the burden of respiratory syncytial virus (RSV), cases are typically classified as healthcare associated if a positive test result occurred after the first 3 days following admission, which may miss healthcare exposures prior to admission. Utilizing an expanded definition of healthcare-associated RSV, we assessed the estimates of disease prevalence. Methods: This study included laboratory-confirmed cases of RSV in adult and pediatric patients admitted to acute-care hospitals in a catchment area of 8 counties in Tennessee identified between October 1, 2016, and April 30, 2019. Surveillance information was abstracted from hospital and state laboratory databases, hospital infection control databases, reportable condition databases, and electronic health records as a part of the Influenza Hospitalization Surveillance Network by the Emerging Infections Program. Cases were defined as healthcare-associated RSV if laboratory confirmation of infection occurred (1) on or after hospital day 4 (ie, “traditional definition”) or (2) between hospital day 0 and 3 in patients transferred from a chronic care facility or with a recent discharge from another acute-care facility in the 7 days preceding the current index admission (ie, “enhanced definition”). The proportion of laboratory-confirmed RSV designated as HA-VRI using both the traditional definition as well as with the added enhanced definition were compared. Results: We identified 900 cases of RSV in hospitalized patients over the study period. Using the traditional definition for HA-VRI, only 41 (4.6%) were deemed healthcare associated. Adding the cases identified using the enhanced definition, an additional 12 cases (1.3%) were noted in patients transferred from a chronic care facility for the current acute-care admission and 17 cases (1.9%) were noted in patients with a prior acute-care admission in the preceding 7 days. Using our expanded definition, the total proportion of healthcare-associated RSV in this cohort was 69 (7.7%) of 900 compared to 13.1% of cases for influenza (Figure 1). Although the burden of HA-VRI due to RSV was less than that of influenza, when stratified by age, the rate increased to 11.7% for those aged 50–64 years and to 10.1% for those aged ≥65 years (Figure 2). Conclusions: RSV infections are often not included in estimates of HA-VRI, but the proportion of cases that are healthcare associated are substantial. Typical surveillance methods likely underestimate the burden of disease related to RSV, especially for those aged ≥50 years.

Funding: No

Disclosures: None

Figure 1.

Figure 2.

Background: Antibiotics are not recommended but are often prescribed for upper respiratory-tract infections (URIs). Prescribers cite patient expectation as a driver of inappropriate antibiotic prescribing; prior literature has demonstrated higher satisfaction scores in patients who receive antibiotics compared to those who do not. We assessed whether veteran satisfaction at URI visits was associated with antibiotic receipt or with reported expectation for antibiotics. Methods: We surveyed veterans with documented URI encounters in the Veterans’ Affairs Tennessee Valley Healthcare System between January 1, 2018, and December 31, 2019. Patients not evaluated in person, with documented dementia, or who died prior to the study start date were excluded. Veterans were asked to recall their URI visit and to complete the Patient Safety Questionnaire (PSQ)-18 (Rand Corporation) and questions assessing antibiotic expectations. The PSQ-18, an 18-item survey that assesses patient satisfaction, uses a 5-point Likert scale (ie, strongly disagree, disagree, uncertain, agree, strongly agree), yielding a composite score of 18–90. Higher scores represent more satisfaction with care. Demographic and visit-specific information were extracted via chart review. We used multivariable linear regression to assess differences in composite PSQ-18 satisfaction scores between those who did and did not receive an antibiotic, adjusted for patient and visit characteristics, and to assess differences in satisfaction scores for those who did and did not report expecting antibiotics, adjusted for antibiotic receipt. Results: We identified 1,435 patients seen for URI at 17 sites. After exclusions, 1,343 veterans were eligible for chart abstraction. After excluding 42 responders who responded after study close or returned blank surveys, the final analytic cohort included 432 (32.2%) of 1,343 responders; 225 (52.1%) received an antibiotic and 207 (47.9%) did not. Mean total satisfaction for veterans who received an antibiotic was 67.8 (SD, ±9.4) compared to 66.7 (SD, ±9.7) for those who did not (Figure 1). Increased total satisfaction was not significantly associated with antibiotic receipt (0.65; 95% CI, −2.0 to 3.3). Most veterans (72.0%) disagreed that visit satisfaction depended on antibiotic receipt. However, only 30.8% reported that they would not expect an antibiotic for URI visits. A significant reduction in total satisfaction (−4.1; 95% CI, −6.3 to −1.9) was associated with expecting compared to not expecting an antibiotic. Conclusions: Our findings suggest that prescribing an antibiotic is not associated with increased veteran satisfaction for URI visits but is associated with expecting an antibiotic. Future work will evaluate methods to change veteran antibiotic expectations.

Funding: No

Disclosures: None

Figure 1.

Background: Healthcare-associated transmission of influenza leads to significant morbidity, mortality, and cost. Most studies classify healthcare-associated viral respiratory infections (HA-VRI) as those with a positive test result after the first 3 days following admission, which does not account for healthcare exposures prior to admission. Utilizing an expanded definition of healthcare-associated influenza, we aimed to improve the estimates of disease prevalence on a population level. Methods: This study included laboratory-confirmed cases of influenza in adult and pediatric patients admitted to any acute-care hospital in a catchment area of 8 counties Tennessee identified between October 1, 2012, and April 30, 2019. Surveillance information was abstracted from hospital and state laboratory databases, hospital infection control practitioner databases, reportable condition databases, and electronic health records as a part of the Influenza Hospitalization Surveillance Network (FluSurv-NET) by the Centers for Disease Control and Prevention (CDC) Emerging Infections Program (EIP). Cases were defined as healthcare-associated influenza laboratory confirmation of infection occurred (1) on or after hospital day 4 (“traditional definition”), or (2) between hospital days 0 and 3 in patients transferred from a chronic care facility or with a recent discharge from another acute-care facility in the 7 days preceding the current index admission (ie, enhanced definition). The proportion of laboratory-confirmed influenza designated as HA-VRI using both the traditional definition as well as with the added enhanced definition were compared. Data were imported into Stata software for analysis. Results: We identified 5,904 cases of laboratory-confirmed influenza in hospitalized patients over the study period. Using the traditional definition for HA-VRI, only 147 (2.5%, seasonal range 1.3%–3.4%) were deemed healthcare associated (Figure 1). Adding the cases identified using the enhanced definition, an additional 317 (5.4%, range 2.3%–6.7%) cases were noted in patients transferred from a chronic care facility for the current acute-care admission and 336 cases (5.7%; range, 4.1%–7.4%) were noted in patients with a prior acute-care facility admission in the preceding 7 days. Using our expanded definition, the total proportion of healthcare-associated influenza in this cohort was 772 of 5,904 (13.1%; range, 10.6%–14.8%). Conclusion: HA-VRI due to influenza is an underrecognized infection in hospitalized patients. Limiting surveillance assessment of this important outcome to just those patients with a positive influenza test after hospital day 3 captured only 19% of possible healthcare-associated influenza infections across 7 influenza seasons. These results suggest that the traditionally used definitions of healthcare-associated influenza underestimate the true burden of cases.

Funding: No

Disclosures: None

Figure 1.