Drug resistance among gram-negative pathogens is a risk factor for inappropriate empiric treatment (IET), which in turn increases the risk for mortality. We explored the impact of carbapenem-resistant Enterobacteriaceae (CRE) on the risk of IET and of IET on outcomes in patients with Enterobacteriaceae infections. We conducted a retrospective cohort study in Premier Perspective database (2009–2013) of 175 US hospitals. We included all adult patients with community-onset culture-positive urinary tract infection (UTI), pneumonia, or sepsis as a principal diagnosis, or as a secondary diagnosis in the setting of respiratory failure, treated with antibiotics within 2 days of admission. We employed regression modeling to compute adjusted association of presence of CRE with risk of receiving IET, and of IET on hospital mortality, length of stay (LOS) and costs. Among 40,137 patients presenting to the hospital with an Enterobacteriaceae UTI, pneumonia or sepsis, 1227 (3.1%) were CRE. In both groups, the majority of the cases were UTI (51.4% CRE and 54.3% non-CRE). Those with CRE were younger (66.6+/−15.3 vs. 69.1+/−15.9 years, p < 0.001), and more likely to be African-American (19.7% vs. 14.0%, p < 0.001) than those with non-CRE. Both chronic (Charlson score 2.0+/−2.0 vs. 1.9+/−2.1, p = 0.009) and acute (by day 2: ICU 56.3% vs. 30.4%, p < 0.001, and mechanical ventilation 35.8% vs. 11.7%, p < 0.001) illness burdens were higher among CRE than non-CRE subjects, respectively. CRE patients were 3× more likely to receive IET than non-CRE (46.5% vs. 11.8%, p < 0.001). In a regression model CRE was a strong predictor of receiving IET (adjusted relative risk ratio 3.95, 95% confidence interval 3.5 to 4.5, p < 0.001). In turn, IET was associated with an adjusted rise in mortality of 12% (95% confidence interval 3% to 23%), and an excess of 5.2 days (95% confidence interval 4.8, 5.6, p < 0.001) LOS and $10,312 (95% confidence interval $9497, $11,126, p < 0.001) in costs. In this large US database, the prevalence of CRE among patients with Enterobacteriaceae UTI, pneumonia or sepsis was comparable to other national estimates. Infection with CRE was associated with a four-fold increased risk of receiving IET, which in turn increased mortality, LOS and costs.

The relationship between multidrug resistance (MDR), inappropriate empiric therapy (IET), and mortality among patients with Acinetobacter baumannii (AB) remains unclear. We examined it using a large U.S. database. We conducted a retrospective cohort study using the Premier Research database (2009–2013) of 175 U.S. hospitals. We included all adult patients admitted with pneumonia or sepsis as their principal diagnosis, or as a secondary diagnosis in the setting of respiratory failure, along with antibiotic administration within 2 days of admission. Only culture-confirmed infections were included. Resistance to at least three classes of antibiotics defined multidrug-resistant AB (MDR-AB). We used logistic regression to compute the adjusted relative risk ratio (RRR) of patients with MDR-AB receiving IET and IET’s impact on mortality. Among 1423 patients with AB infection, 1171 (82.3 %) had MDR-AB. Those with MDR-AB were older (63.7 ± 15.4 vs. 61.0 ± 16.9 years, p = 0.014). Although chronic disease burden did not differ between groups, the MDR-AB group had higher illness severity than those in the non-MDR-AB group (intensive care unit 68.0 % vs. 59.5 %, p < 0.001; mechanical ventilation 56.2 % vs. 42.1 %, p < 0.001). Patients with MDR-AB were more likely to receive IET than those in the non-MDR-AB group (76.2 % MDR-AB vs. 13.8 % non-MDR-AB, p < 0.001). In a regression model, MDR-AB strongly predicted receipt of IET (adjusted RRR 5.5, 95 % CI 4.0–7.7, p < 0.001). IET exposure was associated with higher hospital mortality (adjusted RRR 1.8, 95 % CI 1.4–2.3, p < 0.001). In this large U.S. database, the prevalence of MDR-AB among patients with AB infection was > 80 %. Harboring MDR-AB increased the risk of receiving IET more than fivefold, and IET nearly doubled hospital mortality.

Methicillin-resistant Staphylococcus aureus (MRSA) represents an important pathogen in healthcare-associated pneumonia (HCAP). The concept of HCAP, though, may not perform well as a screening test for MRSA and can lead to overuse of antibiotics. We developed a risk score to identify patients presenting to the hospital with pneumonia unlikely to have MRSA. We identified patients admitted with pneumonia (Apr 2005 – Mar 2009) at 62 hospitals in the US. We only included patients with lab evidence of bacterial infection (e.g., positive respiratory secretions, blood, or pleural cultures or urinary antigen testing). We determined variables independently associated with the presence of MRSA based on logistic regression (two-thirds of cohort) and developed a risk prediction model based on these factors. We validated the model in the remaining population. The cohort included 5975 patients and MRSA was identified in 14%. The final risk score consisted of eight variables and a potential total score of 10. Points were assigned as follows: two for recent hospitalization or ICU admission; one each for age < 30 or > 79 years, prior IV antibiotic exposure, dementia, cerebrovascular disease, female with diabetes, or recent exposure to a nursing home/long term acute care facility/skilled nursing facility. This study shows how the prevalence of MRSA rose with increasing score after stratifying the scores into Low (0 to 1 points), Medium (2 to 5 points) and High (6 or more points) risk. When the score was 0 or 1, the prevalence of MRSA was < 10% while the prevalence of MRSA climbed to > 30% when the score was 6 or greater. MRSA represents a cause of pneumonia presenting to the hospital. This simple risk score identifies patients at low risk for MRSA and in whom anti-MRSA therapy might be withheld.

Biochemical markers of altered mineral metabolism have been associated with increased mortality in end stage renal disease patients. Several studies have demonstrated non-linear (U-shaped or J-shaped) associations between these minerals and mortality, though many researchers have assumed linear relationships in their statistical modeling. This analysis synthesizes the non-linear relationships across studies.