Recent Research

Dabigatran outcomes, PPIs and postoperative pneumonia, and more.


Dabigatran associated with more transfusions, shorter ICU stays, less death than warfarin

Major bleeding patients who took dabigatran instead of warfarin needed more red cell transfusions but less plasma, and they had shorter ICU stays and lower mortality, a study found.

Researchers reviewed bleeding reports from 1,034 individuals with 1,121 major bleeds enrolled in 5 phase III trials that compared dabigatran with warfarin. The 5 trials comprised 27,419 patients with atrial fibrillation (RE-LY) or venous thromboembolism (RE-COVER, RE-COVER II, RE-MEDY, and RESONATE) who were treated for 6 to 36 months. Results were published in the Nov. 19, 2013, Circulation.

Overall among the 5 trials, there were no blood products or other hemostatic agents given for 365 (33%) of the major bleeding events. In terms of receiving blood products or other agents, there were no significant differences between the groups of patients who received dabigatran 110 mg, dabigatran 150 mg, or warfarin. Dabigatran patients more often received red blood cell transfusion alone (without any other blood products, coagulation factors, vitamin K or local hemostatic intervention). Forty-seven percent of patients receiving dabigatran 110 mg and 43% of patients receiving dabigatran 150 mg with bleeds received red blood cell transfusion alone, compared to 20% of warfarin patients.

Based on data from the RE-LY study, ICU stays were shorter for dabigatran patients (1.6 nights, mean) compared to warfarin patients (2.7 nights; P=0.01). There were numerically fewer surgical interventions to stop the bleeding for the dabigatran patients (12.1%) versus the warfarin patients (15%; P=0.17).

Crude mortality at 7 days after the onset of the first major bleeding event among the dabigatran patients was 5.3% versus 8.4% in warfarin patients (P=0.045). At 30 days, it was 9.1% versus 13.0% (P=0.057), respectively. Kaplan-Meier analysis showed a trend to reduced risk for death 30 days after bleeding with both doses of dabigatran compared to warfarin (P=0.052). After adjustment for sex, age, weight, renal function at the time of the bleed and additional antithrombotic therapy, the odds ratio for 30-day mortality in the combined dabigatran treatment groups was 0.66 (95% CI, 0.44 to 1.00; P=0.051) compared to warfarin.

The researchers noted that patients who had a major bleeding event while taking dabigatran were older, had worse renal function and were more often also being treated with aspirin or a nonsteroidal anti-inflammatory agent than those taking warfarin. They wrote, “This implies that when bleeding occurs with dabigatran, the patient is usually at higher risk compared to patients with major bleeding events on warfarin and raises the possibility that some of these bleeds might be avoidable by using a lower dose of dabigatran, as also recommended in some treatment guidelines. Avoidance of concomitant medication with aspirin and non-steroid anti-inflammatory agents might also reduce the risk and severity of bleeding.”

Score predicts likelihood of good outcome after in-hospital CPR

Researchers have developed a simple score, useful in discussing do-not-resuscitate orders, which can identify patients who aren't likely to benefit from a resuscitation attempt if they experience in-hospital cardiac arrest.

Study participants were 51,240 inpatients with an index episode of in-hospital cardiac arrest (IHCA) between 2007 and 2009 in 366 hospitals participating in the Get With the Guidelines-Resuscitation registry. Researchers created several candidate decision models, and used a test data set to choose the model that best classified patients with very low (<1%), low (1% to 3%), average (>3% to 15%) or higher-than-average (>15%) likelihood of surviving with good neurologic status after resuscitation for IHCA. The main outcome was a Cerebral Performance Category score of 1, meaning the patient could lead a normal life and have only minimal neurologic or psychologic deficits, like mild dysphagia. They checked the final model with a validation data set.

The best model used a simple point score based on 13 pre-arrest variables available at the time of admission: neurologically intact or with minimal deficits at admission, major trauma, acute stroke, metastatic or hematologic cancer, septicemia, medical noncardiac diagnosis, hepatic insufficiency, admitted from skilled nursing facility, hypotension or hypoperfusion, renal insufficiency or dialysis, respiratory insufficiency, pneumonia and older age. Results were published in the Nov. 11, 2013, JAMA Internal Medicine.

The resulting clinical prediction rule, the Good Outcome Following Attempted Resuscitation (GO-FAR) score, had a C statistic of 0.78 when applied to the validation set. Overall, the score identified more than a quarter of patients as having low or very low likelihood of surviving to discharge neurologically intact or with minimal deficits after IHCA. Specifically, the score identified 9.4% and 18.9% of patients, respectively, as having very low or low likelihood of a good outcome. Fifty-four percent were identified as having average likelihood of a good outcome and 17.7% as having an above-average likelihood of a good outcome.

The score could be helpful when counseling patients about their “do not resuscitate” (DNR) orders, the authors wrote, adding that past research has found most patients significantly overestimate their post-CPR survival chances. The score could be built into the admission process, they added, and integrated with patient values and expectations to optimize decision making about DNR orders.

Longer delirium time associated with worse cognition a year later

Inpatients who spent more time in delirium had worse cognition and executive function scores at 3 and 12 months, a study found.

In a multicenter, prospective cohort study, researchers examined 821 patients with respiratory failure, cardiogenic shock or septic shock in surgical and medical ICUs. Inpatients were evaluated for delirium (using the Confusion Assessment Method for the ICU) and level of consciousness (using the Richmond Agitation-Sedation Scale) every day until day 30 or hospital discharge.

Separate researchers, who were unaware of the patients' hospital courses, then assessed the patients' global cognition at 3 and 12 months after discharge using the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) and the Trail Making Test, Part B. The researchers used linear regression to examine any associations between duration of delirium, use of analgesic and sedative agents, and outcomes. Results were published in the Oct. 3, 2013, New England Journal of Medicine.

Six percent of patients had cognitive impairment at baseline, and 74% developed delirium while in the hospital. At 3 months, 40% of patients scored 1.5 SDs below the mean in global cognition, similar to scores of patients who have moderate traumatic brain injury. Twenty-six percent of patients had scores 2 SDs below the mean, similar to scores of patients with mild Alzheimer disease.

At 12 months, 34% and 24% of patients had scores at 1.5 and 2 SDs below the mean, respectively. A longer duration of delirium was independently associated with lower global cognition at 3 and 12 months (P=0.001 and P=0.04, respectively), as well as lower executive functioning (P=0.004 and P=0.007 at 3 and 12 months, respectively). Cognitive impairment wasn't associated with analgesic or sedative drug use, and deficits occurred in both older and younger patients.

Though the results were similar to previous studies showing cognitive deficits in ICU survivors, this study is different in that it enrolled a large sample of patients with a broad age range and diverse diagnoses, and detailed data were collected about delirium and sedative exposure as risk factors, the authors noted. A limitation, however, was that researchers couldn't test patients' cognition before their emergent illness, though the researchers did attempt to account for this via exclusions, stratification and statistical adjustments, they added.

Editorialists wrote that the study “unequivocally establishes that critical illness promotes the development of new and clinically important cognitive impairment, regardless of age, burden of coexisting conditions and diagnosis at hospital admission.” The results highlight the important of monitoring for delirium in the ICU and intervening, as well as practicing wake-up and mobility protocols, they wrote.

Patients on contact precautions have more complaints about care

Patients who were under contact precautions during hospitalization had more concerns about the quality of their care, a recent study found.

The prospective cohort study included 528 medical or surgical patients at 1 medical center, 45% of whom were under contact precautions at hospital admission (use of a disposable gown and gloves for all patient contact and segregation in a private or cohorted room). Patients were surveyed on multiple occasions during hospitalization (days 3, 7, and 14) about whether they believed there had been any problems with their care. Results were published in the October 2013 Infection Control and Hospital Epidemiology.

Overall, 20% of the patients reported an issue with their care during hospitalization. Those under contact precautions were more likely to report concerns (odds ratio, 2.04 [95% CI, 1.31 to 3.21]; P<0.01), especially a lack of respect for patient needs and preferences (P=0.001) and poor coordination of care (P=0.02). Responses to the survey's open-ended questions including complaints such as “When people put on the gowns, I feel dirty and alone” and “I didn't get the emotional support I needed.” Even after adjustment for length of stay, education and comorbidities, a significant association remained between contact precautions and patient complaints.

The researchers reached 88 of the study patients (47% of whom had been on contact precautions) by phone after hospitalization to administer the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS). There wasn't a significant association between contact precautions and HCAHPS scores, but a trend toward less satisfaction in the precaution patient group (odds ratio, 1.79 [95% CI, 0.64 to 5.00]; P=0.27).

The study authors observed, “It is unclear whether the patients perceived care that was truly worse or whether the experience of being under contact precautions changed their perception of standard care.” In either case, the authors called for interventions, including patient and clinician education, to ensure that patients under precautions have an equally high-quality experience of care. If this is not possible, the costs and benefits of contact precautions for some patients may need to be re-examined, the authors said.

PPIs associated with risk for postop pneumonia after CABG, study indicates

Proton-pump inhibitors (PPIs) may be associated with a higher risk for postoperative pneumonia after coronary artery bypass grafting (CABG) than H2-receptor antagonists, according to a study.

Researchers performed a retrospective cohort study to determine whether type of stress ulcer prophylaxis after CABG was associated with postoperative pneumonia risk. Patients were included in the study's main analysis if they underwent CABG between 2004 and 2010 on hospital day 3 or later and received stress ulcer prophylaxis on postoperative days 1, 2 or both and were still alive on postoperative day 3. Administrative data were obtained from the Premier Research Database, a complete census of inpatients from about 500 U.S. hospitals. The study's main outcome measure was postoperative pneumonia. The results were published online Sept. 19, 2013, by BMJ.

A total of 21,214 patients were included in the study, 9,830 of whom (46.3%) received PPIs and 11,384 of whom (53.7%) received H2-receptor antagonists after CABG. Four hundred ninety-two of the PPI patients (5.0%) and 487 (4.3%) of the H2-receptor antagonist patients developed postoperative pneumonia. Overall, 101 (10.3%) of the 979 patients who developed pneumonia died in the hospital compared with 232 (1.1%) of the 20,235 patients who did not develop pneumonia. The authors found that the elevated risk for pneumonia with PPIs remained after propensity score adjustment (relative risk, 1.19; 95% CI, 1.03 to 1.38) and that PPI use compared with H2-receptor antagonist use was associated with an increased pneumonia risk of 8.2 cases per 1,000 patients in an instrumental variable analysis.

Among other limitations, the authors acknowledged that the study used an administrative database, relied on coding data rather than clinical data to determine pneumonia occurrence, and excluded certain groups of patients, including those who took acid-suppressing drugs or antibiotics before surgery. However, they concluded that PPIs are associated with an increased risk of postoperative pneumonia in patients who have undergone CABG.

“Given the importance of the complication of postoperative pneumonia in this patient population, clinicians should consider this differential risk when selecting the type of agent used for stress ulcer prophylaxis,” the authors wrote.

Varied surgical readmission rates associated with procedure volume, mortality rate, study finds

Surgical readmission rates for 6 common procedures vary across the U.S. and are associated with hospitals' procedure volume and mortality rate, according to a study.

Researchers used national Medicare data to examine variation in readmission rates after major surgery and to investigate whether such rates are associated with quality of surgical care. Thirty-day readmission rates were calculated for coronary artery bypass grafting, pulmonary lobectomy, endovascular repair of abdominal aortic aneurysm, colectomy and hip replacement. The authors used bivariate and multivariate analysis to determine any association between readmission rates and surgical quality measures, such as procedure volume and mortality. The primary outcome measure was a composite of the risk-adjusted readmission rates for each of the 6 specific procedures at the hospital level at 30 days. The results of the study were published in the Sept. 19 New England Journal of Medicine.

A total of 479,471 patients were discharged from 3,004 hospitals after 1 of the 6 studied procedures. Readmission rates for all 6 procedures ranged from 10.5% to 17.4%. At 30 days, the median risk-adjusted compositor readmission rate was 13.1% (interquartile range, 9.9% to 17.1%). Patients who were readmitted tended to be older and have more comorbid conditions than those who were not. Nonprofit hospitals had lower composite readmission rates than for-profit hospitals, while nonteaching hospitals had lower compositor readmission rates than teaching hospitals (13.1% vs. 13.7% and 12.9% vs. 13.9%, respectively; P<0.001 for both comparisons).

After adjustment for hospital characteristics in a multivariate model, the composite readmission rate was significantly lower in hospitals in the highest surgical-volume quartile compared with the lowest quartile (12.7% vs. 16.8%; P<0.001). In addition, readmission rates were significantly lower in hospitals with the lowest surgical mortality rates compared with the highest (13.3% vs. 14.2%; P<0.001). Adherence to surgical process measures, however, showed only a marginal association with lower readmission rates. These relationships remained consistent when each of the 6 procedures was examined individually.

The authors noted that the clinical significance of their findings is unclear, that they could not account for data not reflected by billing codes, and that their results might not apply to nonelderly adults, among other limitations. Nevertheless, they concluded that readmission rates after surgery were associated with procedure volume and mortality rate, both of which are considered good measures of hospital quality in this area. “Surgical-readmission rates may be a relevant and valid way for policymakers, clinical leaders and patients to measure the surgical performance of U.S. hospitals,” the authors wrote.

Intensive insulin after PCI didn't reduce infarct size

Intensive use of insulin to regulate hyperglycemia in acute coronary syndrome (ACS) patients did not reduce infarct size, and may have caused harm, a study found.

The prospective, open-label study was conducted at a large teaching hospital in the Netherlands. Between 2008 and 2012, 294 patients admitted with ACS and a plasma glucose of 140 to 288 mg/dL were randomized to either intensive glucose management (intravenous insulin therapy targeted at plasma glucose levels of 85 to 110 mg/dL) or conventional management (no insulin therapy unless plasma glucose exceeded 288 mg/dL). Patients with insulin-dependent diabetes before admission were excluded from the study. Results were published in the Nov. 11, 2013, JAMA Internal Medicine.

Percutaneous coronary intervention (PCI) was performed on 93.6% of the study patients, at a median of about 30 minutes after admission. The primary endpoint of the study was high-sensitivity troponin T about 72 hours after admission, with secondary endpoints the area under the curve of creatine kinase, myocardial band, release and myocardial perfusion scintigraphy at 6 weeks. The intensive group (whose median glucose was 112 mg/dL) showed an insignificant decrease in infarct size, as measured by all the endpoints, compared to the conventional group. Although severe hypoglycemia was rare, the intensive group did have a higher incidence of death or spontaneous second myocardial infarction (8 patients [5.7%] vs. 1 [0.7%]; P=0.04).

One possible explanation for the lack of effect is the timing of insulin therapy; it was often delayed until after PCI, a median of 5 hours after symptom onset. It's also possible that elevated plasma glucose was a marker of severity, rather than a causal factor, the authors speculated. They also noted that rapid delivery of PCI could have limited infarct sizes to the point that glucose control didn't provide further benefits. Based on the lack of benefit and possibility of harm observed, “strict, but not too strict, glucose control” seems to be the appropriate strategy, the authors concluded.

They called for additional research to focus on patients with persistently elevated glucose after PCI, as well as the association between intravenous insulin and early second myocardial infarction. An accompanying commentary also suggested a large trial targeting glucose to 140 mg/dL or below, noting that although this study did not show a significant benefit in infarct size, there was a trend favoring intensive glucose management, and other studies show promise with insulin treatment of hyperglycemia in patients with myocardial infarction.

Model predicts bleeding after PCI

A model based on data from the National Cardiovascular Data Registry CathPCI Registry and using an expanded definition of bleeding showed good discrimination predicting bleeding after percutaneous coronary intervention (PCI).

To develop a model of risk factors associated with major bleeding complications occurring within 72 hours after PCI, researchers adapted detailed clinical data from 1,043,759 procedures at 1,142 centers using the CathPCI Registry from February 2008 through April 2011. A full risk score of 33 variables and a simplified risk score of 10 factors were developed from 80% of the cohort and validated from the remaining 20%.

The definition of bleeding was expanded to include events such as intracranial hemorrhage, cardiac tamponade, and transfusions that account for severe anemia. Other criteria included:

  • occurring within 72 hours after PCI or before hospital discharge (whichever occurred first);
  • site-reported arterial access site bleeding, which may be either external or a hematoma >10 cm for femoral access, >5 cm for brachial access, or >2 cm for radial access;
  • retroperitoneal, gastrointestinal or genitourinary bleeding;
  • post-procedure hemoglobin decrease of 3 g/dL in patients with a pre-procedure hemoglobin level ≤16 g/dL; or
  • post-procedure nonbypass surgery-related blood transfusion for patients with a pre-procedure hemoglobin level ≥8 g/dL.

The simplified risk score had 10 factors: ST-segment elevation myocardial infarction, age, body-mass index, previous PCI, chronic kidney disease, shock, cardiac arrest within 24 hours, female sex, hemoglobin, and PCI status (elective, urgent or emergency/salvage). The factors were assigned points and the sum of these factors ranged from 0 (0.90% bleeding risk) to 210 (86% bleeding risk). Results appeared in the September 2013 JACC: Cardiovascular Imaging.

Using the updated definition, the rate of bleeding was 5.8% (n=60,194), which was higher than previously reported rates of 2.4%, the authors noted. The most predictive factors were female sex and shock or salvage PCI, and the least predictive was non-insulin-requiring diabetes. The full bleeding risk model had good discrimination, with a c-index of 0.78 in the development sample and 0.77 in the validation sample. C-indexes for 5 subgroups (ST-elevated myocardial infarction, women, age >70, diabetes and excluding in-hospital coronary artery bypass grafting) ranged from 0.70 to 0.78.

The authors noted that the higher rate of bleeding in this study compared to previous studies reflects the inclusion of complications such as tamponade and transfusions, which generated broader estimates of clinically important bleeding. Another contribution was the number of predictive factors for bleeding events found, such as pre-procedure hemoglobin level, cardiac arrest, shock and clinical status such as salvage procedures, they said.

The researchers wrote, “This model can be used to risk-adjust post-PCI bleeding rates for the centers participating in the CathPCI Registry, identify leaders and laggards, and ultimately improve the safety of PCI by encouraging the adoption of [bleeding avoidance strategies] at centers that have higher than-expected risk-adjusted bleeding rates.”

Hospital-acquired anemia associated with higher death rates

Hospital-acquired anemia (HAA) is common and appears to be associated with increased mortality rates, according to a study.

Researchers analyzed hospitalizations at 1 academic medical center and 9 community hospitals in the Cleveland Clinic Health System from Jan. 1, 2009, to Aug. 31, 2011, to evaluate the prevalence of HAA and to determine whether an association existed among HAA and increased mortality, length of stay (LOS) or total hospital charges.

Mild HAA was defined as hemoglobin levels >11 g/dL and <12 g/dL in women and >11 g/dL and <13 g/dL in men, moderate HAA was defined as a hemoglobin level of 9.1 to ≤11.0 g/dL, and severe HAA was defined as a hemoglobin level ≤9.0 g/dL. Information on demographics, comorbid conditions and outcomes was obtained from administrative data, while data on hemoglobin values were obtained from the electronic medical record. The study appeared in the September 2013 Journal of Hospital Medicine.

Among the 188,447 hospitalizations included in the analysis, 139,807 patients (74%) developed HAA. HAA was mild in 29% of cases, moderate in 41% and severe in 30%. Patients were more likely to develop HAA if they were older and had more comorbid conditions.

Risk-adjusted odds ratios for in-hospital mortality in patients with HAA were 1.0 for mild HAA (P=0.8), 1.51 for moderate HAA (P<0.001) and 3.28 for severe HAA (P<0.001). Patients with HAA also had higher risk-adjusted relative mean LOS and hospital charges compared with patients who didn't have HAA; risk-adjusted odds ratios were 1.08, 1.28 and 1.8 for LOS and 1.06, 1.18 and 1.80 for hospital charges with mild, moderate and severe HAA, respectively (P<0.001 for all comparisons).

These summaries come from ACP HospitalistWeekly, an e-newsletter provided every Wednesday by ACP Hospitalist. To receive it, call Customer Service at 800-543-1546, ext. 2600, or direct at 215-351-2600 (M-F, 9 a.m. to 5 p.m. EST); email; or subscribe online.