Heart failure risk score, step-down therapy

Summaries from ACP Hospitalist Weekly.

Risk score appears valid for predicting mortality in ED patients with acute heart failure

A Spanish risk score to predict 30-day mortality in patients presenting to the ED with acute heart failure was externally validated in a recent prospective cohort study.

Image by Getty Images
Image by Getty Images

Researchers used data from a study of adult patients in Switzerland who presented to the ED with dyspnea at two university hospitals and two nonacademic tertiary care centers, including those who had an adjudicated final diagnosis of heart failure, as determined by two independent physicians, and 12 of the 13 variables needed for calculation of the MEESSI-AHF (Multiple Estimation of risk based on the Emergency department Spanish Score In patients with Acute Heart Failure) score. Although the full model for the score includes 13 variables collected at ED admission, the current study omitted one, the Barthel Index, because it was not recorded prospectively.

Patient outcomes were determined by self-report when patients were contacted by telephone or in writing three, 12, and 24 months after discharge. Data on death during follow-up were obtained from hospital medical records, patients' general practitioners, and the national mortality registry. The study results were published Jan. 29 by Annals of Internal Medicine and appeared in the Feb. 19 issue.

Overall, 1,572 patients had adjudicated heart failure, and of these, 1,247 had data available on the 12 variables needed to calculate the MEESSI-AHF score and were eligible for the primary analysis. One hundred two of the 1,247 patients (8.2%) died within 30 days. Patients were stratified into six risk groups according to predicted probability of death. The MEESSI-AHF score had excellent discrimination for predicting 30-day mortality (c-statistic, 0.80), with a steep gradient in 30-day mortality over the six predefined risk groups for assessment of cumulative mortality.

The lowest-risk group had a 0.5% to 2.1% predicted probability of death, while the highest-risk group had a predicted probability of 25.8% to 89.8%. In the former group, 0% of patients died within 30 days of presentation, and in the latter, 28.5% died. While the score overestimated risk in the high-risk groups, good concordance was found between predicted risk and observed outcome after adjustment of the intercept (P=0.23). The researchers confirmed their findings in sensitivity analyses with multiple imputations for missing values in the overall cohort of 1,572 patients.

The entire MEESSI-AHF score was not used in this validation study, and the results apply only to clinically stable patients who present to the ED with acute heart failure, the authors noted. However, despite these and other limitations, they concluded that the MEESSI-AHF risk score can predict 30-day mortality in acute heart failure with excellent discrimination. They noted that the score has several advantages, including its development with nonadministrative data, its inclusion of palliative patients, and its derivation from a large population-based cohort.

The intercept for the score should be recalibrated for specific populations by using its online calculator, the authors recommended, noting that recalibration could also be necessary in cases of novel therapies and consecutive improved outcomes. “Clinical use of the MEESSI-AHF score could improve patient and economic outcomes by identifying high-risk patients who can benefit from prompt and aggressive in-hospital therapy and triaging low-risk patients who can be safely discharged from the ED without exposure to potential hazards, such as nosocomial infections and delirium,” the authors wrote.

Early oral step-down therapy did not increase 30-day mortality rates for Enterobacteriaceae bacteremia

Early oral step-down therapy for patients with Enterobacteriaceae bacteremia was associated with similar clinical outcomes as continued parenteral therapy, but switching to oral therapy might lead to faster discharge, a recent study found.

Researchers retrospectively assessed a cohort of 2,161 individuals (median age, 59 years) with monomicrobial Enterobacteriaceae bacteremia who were hospitalized at three U.S. academic medical centers from Jan. 1, 2008, through Dec. 31, 2014. They compared patients who received continued parenteral therapy versus oral step-down therapy within five days of treatment to evaluate differences in rates of 30-day all-cause mortality and bacteremia recurrence. Eligible patients had appropriate source control measures in place, active antibiotic therapy from day one until discontinuation, clinical response by day five, availability of an active oral antibiotic option, and the ability to consume other oral medications or food. Results were published online on Jan. 22 by JAMA Internal Medicine.

After propensity score matching, there were 1,478 participants remaining, 739 in each study arm. The most common sources of bacteremia were urine (40.2%), gastrointestinal tract (20.1%), and central line (18.4%). Overall, 97 (13.1%) patients in the oral step-down group died within 30 days, compared to 99 (13.4%) patients in the IV group (hazard ratio, 1.03; 95% CI, 0.82 to 1.30). There was also no significant difference between groups in recurrence of bacteremia within 30 days (6 [0.8%] patients in the IV group vs. 4 [0.5%] in the oral group; hazard ratio, 0.82 [95% CI, 0.33 to 2.01]).

A significant difference was found, however, in time to discharge from the hospital. Patients who transitioned to early oral step-down therapy were discharged an average of two days (interquartile range [IQR], 1 to 6 days) sooner than those who continued IV therapy (5 days [IQR, 3 to 8 days] vs. 7 days [IQR, 4 to 14 days]; P<0.001).

The authors noted limitations of the study, such as its retrospective design and the fact that stringent propensity matching led to the exclusion of a large number of patients. “Until a clinical trial is performed, our findings suggest that oral step-down therapy is not associated with inferior clinical outcomes for patients with Enterobacteriaceae bacteremia who have received appropriate source control and demonstrated an appropriate clinical response compared with patients who continue to receive IV therapy for the duration of their treatment course,” they concluded.

Influenza vaccination administration during hospitalization appears to be safe

Influenza vaccination during hospitalization was not associated with an increased risk of readmissions, outpatient or ED visits, fever, or clinical evaluations for infection in the week after discharge, a recent study found.

The retrospective cohort study included patients ages six months and older who were hospitalized at a Kaiser Permanente Southern California facility. Researchers identified all influenza vaccinations administered at the hospital during the period of Aug. 1 to April 30 for influenza seasons 2011 to 2012, 2012 to 2013, and 2013 to 2014. Using propensity score analyses with inverse probability of treatment weighting, they compared patients who received an influenza vaccination during their hospitalization with those who were never vaccinated that season or were vaccinated at other times.

Image by Getty Images
Image by Getty Images

The study assessed 290,149 hospitalizations among 255,737 patients (median age, 58.9 years; 62.1% female) for the following outcomes in the seven days following discharge: rates of outpatient and ED visits, readmissions, fever (defined as temperature ≥100.4 °F or 38 °C), and clinical laboratory evaluations for infection (e.g., urine, blood, or wound culture; complete blood cell count). Results were published online on Jan. 8 by Mayo Clinic Proceedings and appeared in the March issue.

Overall, 28% (41,733 of 147,670) of patients who were not already vaccinated before hospitalization were vaccinated for influenza before discharge. After discharge, 26% (27,128 of 105,937) of the remaining unvaccinated patients received the vaccine.

In adjusted analyses, there was no increased risk associated with vaccination during hospitalization for any of the study outcomes. After adjustment, there were decreases in readmissions (relative risk [RR], 0.88; 95% CI, 0.83 to 0.95; P<0.001), outpatient visits (RR, 0.97; 95% CI, 0.95 to 0.99; P=0.008), fever (RR, 0.80; 95% CI, 0.68 to 0.93; P=0.005), and clinical evaluations for infection (RR, 0.95; 95% CI, 0.92 to 0.98; P=0.002) among patients vaccinated during hospitalization compared with those who were never vaccinated or who were vaccinated at other times.

The study authors noted limitations, such as the fact that they did not distinguish between planned and unplanned readmissions or outpatient visits. “There remains a prominent care gap in the missed opportunities for influenza vaccination in the inpatient setting,” they concluded. “However, the data demonstrating effectiveness of influenza vaccination and our data about the safety of inpatient influenza vaccination provide assurance that the closure of this care gap is a reasonable strategy to reduce the burden of morbidity and mortality due to influenza.”

ICU clinician education increased guideline compliance, decreased ICU brain dysfunction

ICUs that implemented an education program on delirium successfully improved compliance with recommendations and reduced the duration of delirium among their patients, a recent study showed.

The study included 3,930 medical and surgical patients treated in ICUs in the Netherlands between April 1, 2012, and February 1, 2015. The primary outcome was adherence to recommendations on managing ICU delirium from the American College of Critical Care Medicine's 2013 clinical practice guidelines. The study included a baseline measurement phase, a second phase in which all nurses and physicians were required to complete an e-learning program on delirium screening, and a third phase in which the clinicians completed a second e-learning program focused on the guidelines. Results were published by Critical Care Medicine on Jan. 2 and appeared in the March issue.

The study authors found “modest, although, significant” improvements in almost all of the performance metrics of guideline compliance they studied: delirium screening (from 35% to 96%; P<0.001), use of benzodiazepines for continuous sedation (from 36% to 17%; P<0.001), light sedation of ventilated patients (from 55% to 61%; P<0.001), physiotherapy (from 21% to 48%; P<0.001), and early mobilization (from 10% to 19%; P<0.001). Use of an analgesia-first sedation approach did not improve, and use of sedation assessment did not increase significantly from the second phase to the third.

There were significant improvements in the study's secondary outcome of brain dysfunction: The mean delirium duration decreased from 5.6 to 3.3 days (–2.2 days; 95% CI, –3.2 to –1.3; P<0.001), and coma days decreased from 14% to 9% (risk ratio, 0.5; 95% CI, 0.4 to 0.6; P<0.001). However, other clinical outcome measures, including length of mechanical ventilation, length of ICU stay, and hospital mortality, did not change.

“Our findings provide empirical support for the differential efficacy of the guideline bundle elements in a real-life setting and provide lessons for optimization of guideline implementation programs,” the authors said. They noted that they chose not to focus their intervention heavily on screening for spontaneous awakening and spontaneous breathing trials on the assumption that such improvements would follow naturally and that the study's results disproved this assumption.

The authors believe this to be the largest study of its kind conducted outside the U.S., with a mix of ICU types that should make the findings widely applicable. It also shows “the feasibility of staggered versus simultaneous implementation of bundle elements” using local program champions, the authors said. Limitations of the study include the possibility of the Hawthorne effect and the uncertain accuracy of duration of delirium measurements.

Coding change may explain half of reduction in readmissions seen under the HRRP

Previous analyses of Medicare's Hospital Readmissions Reduction Program (HRRP) may have overstated the program's effects on readmissions, according to a recent study.

The study authors constructed three samples of admissions: one of admissions for HRRP-targeted conditions to targeted hospitals (7,049,806 index admissions to 3,350 hospitals), one of admissions for nontargeted conditions to targeted hospitals (40,148,231 index admissions to 3,467 hospitals), and one of admissions for targeted conditions to nontargeted hospitals (429,072 index admissions to 1,115 critical access hospitals). Results were published in the January Health Affairs.

The study also looked at a change in electronic transaction standards that happened just after the HRRP was implemented, allowing hospitals to report more diagnosis codes per patient. In November 2010, 10 diagnoses were the maximum that could be submitted and 81% of admissions reported nine or 10 diagnoses. After the standards change, in January 2011, 70% of admissions reported 11 or more diagnoses, while only 15% reported nine or 10 diagnoses. Accounting for these revised standards reduced the decline in risk-adjusted readmission rates for targeted conditions under the HRRP by 48%, the study authors found.

“The electronic transaction standards update does not explain the entire decline in risk-adjusted readmission rates following implementation of the HRRP,” they wrote. To explain the rest, they turned to the comparison samples, where they found that declines in readmissions for targeted conditions at targeted hospitals “were statistically indistinguishable from declines in two control samples,” they said. The results suggest that either the HRRP had no effect on readmissions or it led to a systemwide reduction in readmissions, affecting hospitals and conditions whether or not they were targeted, according to the authors. “Distinguishing between these conclusions remains an important topic for research,” they said.

The authors noted that it's important to determine the HRRP's effects, because such pay-for-performance programs carry potential downsides, including providing motivation to discharge patients from the ED or put them on observation status rather than admit them, and creating instability in hospital funding, such as penalties for safety net hospitals that have already been associated with the HRRP. “In a successful pay-for-performance program, these two potential downsides must be more than made up for by robust improvements in performance. Our study suggests that any salutary effects of the HRRP are smaller than earlier estimates have suggested,” the authors concluded.

Mortality rates not affected by reductions in inpatient transfusions for moderate anemia

Limiting blood transfusions for inpatients with moderate anemia did not increase transfusions, mortality, or readmissions in the six months after hospital discharge, a recent study found.

The retrospective cohort study included 445,371 adults who had 801,261 hospitalizations at 21 northern California hospitals from January 2010 through December 2014. Among these patients, the prevalence of moderate anemia (hemoglobin levels between 7 and 10 g/dL) at hospital discharge increased from 20% in 2010 to 25% in 2014 (P<0.001), and the rate of red blood cell transfusion declined by 28% (from 39.8 to 28.5 red blood cell units per 1,000 patients; P<0.001). The proportion of patients whose moderate anemia had resolved within six months of hospital discharge decreased from 42% to 34% (P<0.001).

However, rates of red blood cell transfusion and rehospitalization within six months of hospital discharge also decreased in the same period, from 19% to 17% and 37% to 33%, respectively (P<0.001 for both). Among the patients with moderate anemia at discharge, the adjusted six-month mortality rate also decreased, from 16.1% to 15.6% (P=0.004). The mortality trend was similar among patients without moderate anemia. The study was published by Annals of Internal Medicine on Dec. 18, 2018, and appeared in the Jan. 15 issue.

The study authors concluded that the results support the safety of practice recommendations to limit red blood cell transfusions and tolerate anemia during and after hospitalization. They noted that previous research has shown the safety of moderate anemia during hospitalization but that there was concern about the potential for adverse outcomes after discharge. “These growing data provide reassurance to hospital clinicians who must decide whether to discharge patients with anemia,” the study authors said. They did caution that anemia could still affect patients' quality of life and called for research on this outcome, as well as research on treatments for anemia other than transfusions.

An accompanying editorial expressed even more strongly that the focus should be on preventing anemia and treating it with methods other than transfusion. “Given the ample evidence on harms of allogeneic blood transfusions and anemia, when the decision is reduced to choosing between transfusion and ‘tolerance’ (that is, acceptance) of anemia, we are left to choose the lesser of 2 evils,” the editorial said. It called for intensified multidisciplinary patient blood management programs before, during, and after surgery as well as for nonsurgical patients.

Outcomes similar with one or two weeks of antibiotics for gram-negative bacteremia

For stabilized patients with gram-negative bacteremia, seven days of antibiotic therapy worked as well as 14 days, a recent trial found.

The randomized trial included 604 patients with gram-negative bacteremia who were hospitalized in three centers in Israel and Italy between January 2013 and August 2017. Patients with an uncontrolled focus of infection were excluded. The majority (68%) had a urinary source of infection, and 90% of the causative pathogens were Enterobacteriaceae. After patients were afebrile and hemodynamically stable for at least 48 hours, they were randomized to receive covering antibiotic therapy for either seven (n=306) or 14 days (n=298). Results were published by Clinical Infectious Diseases on Dec. 11, 2018.

The study's primary outcome was a composite of all-cause mortality; relapse, suppurative, or distant complications; and readmission or extended hospitalization (>14 days) within 90 days. It did not differ significantly between the groups (45.8% in the seven-day group vs. 48.3% in the 14-day group; risk difference, −2.6% [95% CI, −10.5% to 5.3%]). The groups were similar on all other outcomes and adverse events, except that the seven-day group reported returning to baseline functional status sooner.

The authors noted that shorter antibiotic treatment would be expected to reduce adverse events but that this might not have been seen due to low rates of Clostridium difficile infection and other adverse events, as well as the difficulty of monitoring for effects on resistance in a relatively short trial. The finding of a more rapid return to baseline function was based on patients' subjective assessment, which might be affected by their perception of illness while taking antibiotics, the authors also said.

The findings apply only to patients who survive gram-negative bacteremia and achieve hemodynamic stability for at least 48 hours before day 7 with no uncontrolled source of infection, the authors noted. Because Enterobacteriaceae was the primary pathogen, there is also limited applicability of the results to infections with Pseudomonas aeruginosa or Acinetobacter baumannii.

There are currently limited data on the optimal duration of therapy for gram-negative bacteremia, according to the authors. A couple of studies have found that shorter durations didn't significantly affect rates of mortality or relapse, but a previous trial similar to this one found increased treatment failure with fewer than 10 days of antibiotics, both overall and in the subgroup of patients with urinary tract infection, the study authors noted.

Veterans Health Administration hospitals outperformed other facilities on mortality, safety measures

Across hospital referral regions, care at Veterans Health Administration (VHA) hospitals was significantly better than care at non-VHA hospitals for 14 of 15 outcome measures, a recent analysis found.

Image by Getty Images
Image by Getty Images

Researchers sought to use Hospital Compare data as a veteran would if trying to find the best local care. They identified 15 outcome measures reported by both VHA and non-VHA hospitals, which included 30-day risk-adjusted mortality rates for acute myocardial infarction, chronic obstructive pulmonary disease, heart failure, and pneumonia, plus 11 additional patient safety indicators (e.g., pressure sores, postoperative respiratory failure, bloodstream infection after surgery). Of 306 potential hospital referral regions defined by The Dartmouth Atlas of Health Care, the researchers analyzed 121 regions in which at least one VHA hospital and one non-VHA hospital reported one or more measures.

For each measure and region, the researchers used the t-test to compare mean VHA scores with mean non-VHA scores to determine whether the score of a VHA hospital was the best, above average, or the worst among all hospitals in the region. Results were published as a research letter on Dec. 11, 2018, by Annals of Internal Medicine and appeared in the March 19 issue.

Across all regions, VHA care was significantly better than non-VHA care for 14 of the 15 measures (the nonsignificant exception being pressure sores). A VHA hospital provided the best care in at least 50% of the regions for nine measures, above-average care in at least 67% of the regions for 14 measures, and the worst care in fewer than 15% of the regions for 14 measures.

The authors, who are both former VHA employees, noted limitations of the study, such as its cross-sectional nature and the possibility that VHA and non-VHA hospitals report data differently to Hospital Compare. The results “may indicate that the VHA generally provides truly excellent care; if so, we believe that efforts to outsource VHA care to non-VHA settings solely for patient convenience should be reconsidered,” they wrote.