qSOFA for mortality prediction, preoperative breathing exercises, and more

Summaries from ACP Hospitalist Weekly.


For mortality prediction, qSOFA has poor sensitivity, moderate specificity, review finds

The quick Sequential Organ Failure Assessment (qSOFA) score had poor sensitivity and moderate specificity for predicting short-term mortality in hospital and ED patients with suspected infection, according to a review.

The systematic review and meta-analysis included 38 studies (n=385,333) that used qSOFA to predict in-hospital, 28-day, or 30-day mortality of patients with suspected infection in the ICU, ED, or hospital wards. The prognostic accuracy of qSOFA was compared with the systemic inflammatory response syndrome (SIRS) criteria. Results were published Feb. 6 by Annals of Internal Medicine and appeared in the Feb. 20 issue.

The pooled sensitivity for qSOFA was 60.8% (95% CI, 51.4% to 69.4%), and specificity was 72.0% (95% CI, 63.4% to 79.2%). In comparison, the SIRS criteria had a pooled sensitivity of 88.1% (95% CI, 82.3% to 92.1%) and specificity of 25.8% (95% CI, 17.1% to 36.9%). The pooled sensitivity of qSOFA was higher in the ICU population (87.2%; 95% CI, 75.8% to 93.7%) than the non-ICU population (51.2%; 95% CI, 43.6% to 58.7%), while the pooled specificity had the opposite pattern—higher in the non-ICU population (79.6%; 95% CI, 73.3% to 84.7%) than the ICU population (33.3%; 95% CI, 23.8% to 44.4%).

The study authors concluded that their findings support the use of SIRS for screening and initiating treatment for sepsis. “This extensive review of the literature provides the best available assessment of the prognostic accuracy of the novel qSOFA tool,” they wrote. They noted that the Sepsis-3 Task Force did not intend to replace the SIRS criteria when it proposed qSOFA but intended the latter as “an early warning risk stratification tool for identification and escalation of care in patients at high risk for death.”

It's unclear why the qSOFA was far more sensitive but far less specific in the ICU than outside it, but understanding how qSOFA performs differently in these settings is critical, the authors said. They noted several limitations of the review, primarily relating to the quality of the included studies, including potential lack of blinding, very specific patient populations, and differing definitions of suspected infection.

An accompanying editorial described qSOFA as “by far the most controversial and, alas, the most misunderstood aspect” of Sepsis-3. “The utility of qSOFA as a clinical decision support tool needs to be established. However, an advantage over the SIRS criteria is that qSOFA does not require laboratory tests and can be used quickly and repeatedly,” the editorialists wrote. They recommended that the two screening tools be viewed as complementary rather than competing.

Training in breathing exercises before upper abdominal surgery reduced postoperative pulmonary complications

Brief training in self-directed breathing exercises before upper abdominal surgery may help patients avoid postoperative pulmonary complications, a randomized trial found.

Researchers enrolled adults who were scheduled to receive elective major open upper abdominal surgery (e.g., colectomy, other bowel resection, nephrectomy) at three hospitals in Australia and New Zealand. Overall, 432 patients completed the trial. Within six weeks before surgery, the control group received an information booklet that prescribed postoperative breathing exercises (n=214), while the intervention group received both the booklet and a 30-minute training session with a physiotherapist (n=218), which involved coaching in the breathing exercises, education about pulmonary complications, and an individualized risk assessment.

The primary outcome was incidence of a postoperative pulmonary complication during the patient's hospital stay (up to 14 days). Secondary outcomes included hospital-acquired pneumonia, length of stay, ICU utilization, and hospital costs. Results were published online on Jan. 24 by The BMJ.

The intervention group had a lower incidence of postoperative pulmonary complications than the control group (12% vs. 27%), an absolute risk reduction of 15% (95% CI, 7% to 22%; P<0.001). The incidence remained halved after adjustment for age, respiratory comorbidity, and type of surgical procedure (adjusted hazard ratio, 0.48 [95% CI, 0.30 to 0.75]; P=0.001), and the number needed to treat to prevent one pulmonary complication was 7 (95% CI, 5 to 14). The effect of the intervention was more pronounced in patients having colorectal surgery, in patients younger than age 65 years, in men, and in those who received training from an experienced physiotherapist.

The incidence of hospital-acquired pneumonia was also lower in the intervention group than in the control group (8% vs. 20%; adjusted hazard ratio, 0.45 [95% CI, 0.26 to 0.78]; P=0.005), for a number needed to treat of 9 (95% CI, 6 to 21). There were no significant differences between groups in other secondary outcomes. The study authors noted limitations, such as imbalances in baseline characteristics between groups and limited generalizability.

The intervention has minimal potential to harm and can be provided within existing hospital clinics that patients must attend before surgery, the authors noted. “These results are directly applicable to the tens of millions of patients listed for elective major abdominal surgery worldwide,” they concluded. “This service could be considered for all patients awaiting upper abdominal surgery.”

Prognostic value of acute-to-chronic glycemic ratio better than glycemia at admission for AMI patients

For patients with acute myocardial infarction (AMI), the acute-to-chronic (A/C) glycemic ratio predicts in-hospital morbidity and mortality better than glycemia at admission, according to a study.

In an observational study at a single facility in Italy, researchers prospectively measured glycemia at admission and calculated A/C glycemic ratio in consecutive patients with AMI to determine which indicator had better prognostic value. Patients were considered to have diabetes if diabetes or use of antidiabetes treatment was noted in the admission history, and unknown diabetes was diagnosed if patients had an HbA1c of 6.5% or greater at admission and no history of the disease. The primary end point of the study was a composite of in-hospital mortality, acute pulmonary edema, and cardiogenic shock. The study results were published Jan. 30 by Diabetes Care.

Overall, 1,553 patients were included in the study, 747 with ST-elevation MI (STEMI) and 806 with non-ST-elevation MI (NSTEMI). Mean age was 67 years, and most patients (74%) were men. Four hundred seventeen patients (27%) had diabetes, 233 (15%) had acute hyperglycemia according to a glucose cutoff of 198 mg/dL, and 583 (37%) had acute hyperglycemia according to a glucose cutoff of 144 mg/dL. Rate of the primary end point increased along with tertiles of A/C glycemic ratio (P<0.0001 for trend), and a parallel increase was seen in troponin I peak value (P<0.0001). A/C glycemic ratio remained an independent predictor of both the primary end point and troponin I peak value in multivariable analysis, with and without adjustment for major confounders. Reclassification analyses were performed and found that A/C glycemic ratio had the best prognostic power for the primary end point in the overall study population (12% net reclassification improvement; P=0.003) and especially in patients with diabetes (27% net reclassification improvement; P<0.0001).

The authors noted that most of the patients in their study were treated with percutaneous coronary intervention and that their results may not apply to patients with AMI who do not undergo that procedure. The study may also have been limited by its observational design and by the possibility that average chronic glucose level was underestimated in patients who had low hemoglobin levels at admission, among other factors, they said. However, they concluded that the A/C glycemic ratio appears to be closely associated with morbidity and mortality rates during hospitalization in patients with AMI. “Use of the A/C glycemic ratio may be particularly valuable in patients with diabetes with chronically elevated glycemic levels because it may identify true stress hyperglycemia, which has been associated with larger infarct size and worse in-hospital outcome,” the authors wrote.

Fewer than 1% of ED patients with syncope had pulmonary embolism in retrospective analysis

Very few patients who present to an ED with syncope are found to have a pulmonary embolism (PE), according to a large, retrospective study.

The observational study used administrative data from databases in Canada, Denmark, Italy, and the U.S. It included 1.6 million adults who presented to an ED for syncope between Jan. 1, 2000, and Sept. 30, 2016. The primary outcome was prevalence of PE at ED or hospital discharge. Results were published by JAMA Internal Medicine on Jan. 29 and appeared in the March issue.

Across the databases, the prevalence of PE ranged from 0.06% (95% CI, 0.05% to 0.06%) to 0.55% (95% CI, 0.50% to 0.61%) for all patients. For patients who were hospitalized, rates of PE before discharge ranged from 0.15% (95% CI, 0.14% to 0.16%) to 2.10% (95% CI, 1.84% to 2.39%). The researchers also conducted two sensitivity analyses. The first looked at prevalence of all PE at 90 days to identify all possibly unrecognized PEs and found it to range from 0.14% to 0.83% for all patients and from 0.35% to 2.63% for hospitalized patients. The second sensitivity analysis looked at prevalence of any venous thromboembolism at 90 days, assuming a worst-case scenario that every identified venous thromboembolism was a PE, and found that it ranged from 0.30% to 1.37% for all patients and from 0.75% to 3.86% for hospitalized patients.

The study authors noted that PE had been considered an uncommon cause of syncope until the Pulmonary Embolism in Syncope Italian Trial (PESIT) found the prevalence to be as high as 17% in patients hospitalized for first syncope and 3.7% in patients presenting to the ED with syncope. The current analysis, in contrast, is consistent with most studies that have recruited patients with syncope in the ED and found a PE prevalence of less than 1.5%. Study authors speculated that the higher rates in PESIT might be due to use of a structured algorithm, which increased identification of subsegmental PE and false-positive results, or the inclusion of only hospitalized patients, who have higher PE risk.

“Our study has relevant clinical implications; our findings discourage the adoption of a routine protocol for excluding PE in all patients presenting to the ED with syncope,” they wrote. Testing all syncope patients for PE could lead to unnecessary risks of radiation exposure, contrast allergies, and anticoagulation, as well as being expensive and time-consuming, the authors said. They concluded that PE should be considered as a differential diagnosis in every patient who presents with syncope but that not all such patients warrant evaluation for it.

After ICH, patients on NOACs appear more likely to survive to discharge than those on warfarin

Patients taking non-vitamin K antagonist oral anticoagulants (NOACs) were less likely to die in the hospital after an intracerebral hemorrhage (ICH) than those taking warfarin, a study found.

Researchers conducted a retrospective cohort study of 141,311 patients with ICH admitted from October 2013 to December 2016 to any of 1,662 hospitals participating in the Get With The Guidelines–Stroke registry. In the seven days before ICH, 10.6% of the patients were taking warfarin and 3.5% were taking NOACs; 28.0% of these patients were taking concomitant antiplatelets and 4.1% were on dual antiplatelet therapy. Results were published online by JAMA on Jan. 25 and appeared in the Feb. 6 issue.

Unadjusted in-hospital mortality rates were 32.6% for patients on warfarin, 26.5% for patients on NOACs, and 22.5% for patients not on OAC, with no significant differences in acute ICH stroke severity (measured by the National Institutes of Health Stroke Scale) across the three groups. In direct comparisons with the no-OAC group, in-hospital mortality was higher in both the warfarin group (adjusted risk difference [RD], 9.0%; adjusted odds ratio [OR], 1.62; 97.5% CI, 1.53 to 1.71) and the NOAC group (RD, 3.3%; OR, 1.21; 97.5% CI, 1.11 to 1.32).

In-hospital mortality was also significantly lower in the NOAC group compared to the warfarin group (RD, −5.7%; OR, 0.75; 97.5% CI, 0.69 to 0.81). This mortality difference was numerically greater among patients on dual antiplatelet agents (32.7% vs. 47.1%; RD, −15.0%; OR, 0.50; 97.5% CI, 0.29 to 0.86) than in those not on antiplatelets (26.4% vs. 31.7%; RD, −5.0%; OR, 0.77; 97.5% CI, 0.70 to 0.85), although the interaction was not statistically significant.

To the authors' knowledge, this study is the largest reporting clinical experience with anticoagulation-related ICH. Based on the results, they concluded that prior use of NOACs or warfarin was associated with higher in-hospital mortality compared with no oral anticoagulation and that NOACs were associated with lower risk than warfarin. This suggests “that NOACs may be a better option than warfarin,” the authors said. The study had relatively few patients on concomitant NOAC and antiplatelet therapy, but results support that NOACs may also have a better safety profile for such patients, the authors said. They noted that the majority of study patients (63%) were treated before the approval of a reversal agent for any NOAC.

The study had several limitations, including the risk of residual or unmeasured confounding, which might have been particularly likely in comparisons with patients not taking any oral anticoagulant. The authors did perform adjustments, but patients with prior use of warfarin or NOACs were older and had higher prevalence of atrial fibrillation and prior stroke than those who had taken neither type of drug.

Continuous hydrocortisone did not reduce mortality from septic shock

Among patients with septic shock undergoing mechanical ventilation, a continuous infusion of hydrocortisone did not result in lower 90-day mortality than placebo, a study found.

To specifically target a population with high requirements for vital organ support (use of mechanical ventilation and at least four hours of vasopressor therapy before randomization) and a substantial risk of death, researchers randomly assigned patients with septic shock to receive 200 mg of hydrocortisone per day or placebo for seven days or until death or discharge from the ICU, whichever came first. Results were published online by the New England Journal of Medicine on Jan. 19 and appeared in the March 1 issue.

From March 2013 through April 2017, 3,800 patients were enrolled. Status with respect to the primary outcome, death from any cause at 90 days, was determined in 3,658 patients (1,832 of whom had been assigned to the hydrocortisone group and 1,826 of whom had been assigned to the placebo group). At 90 days, 27.9% of patients in the hydrocortisone group had died, compared to 28.8% in the placebo group (odds ratio, 0.95; 95% CI, 0.82 to 1.10; P=0.50). The effect of the trial regimen was similar in six prespecified subgroups. Patients assigned to receive hydrocortisone had faster resolution of shock than those on placebo (median duration, 3 days [interquartile range, 2 to 5 days] vs. 4 days [interquartile range, 2 to 9 days]; hazard ratio [HR], 1.32; 95% CI, 1.23 to 1.41; P<0.001). Patients in the hydrocortisone group also had a shorter duration of the initial episode of mechanical ventilation than those in the placebo group (median, 6 days [interquartile range, 3 to 18 days] vs. 7 days [interquartile range, 3 to 24 days]; HR, 1.13; 95% CI, 1.05 to 1.22; P<0.001). However, when episodes of recurrence of ventilation were taken into account, no significant differences were seen in the number of days alive and free from mechanical ventilation. Blood transfusion was less common in the hydrocortisone group than in the placebo group (37.0% vs. 41.7%; odds ratio, 0.82; 95% CI, 0.72 to 0.94; P=0.004).

No significant between-group differences were seen in mortality at 28 days, rate of shock recurrence, number of days alive and out of the ICU, number of days alive and out of the hospital, recurrence of mechanical ventilation, rate of renal replacement therapy, and incidence of new-onset bacteremia or fungemia.

The researchers wrote, “Although we did not observe a significant difference between the hydrocortisone group and the placebo group with regard to 90-day mortality, some secondary outcomes were better in the group that received the active treatment. Our observation of the hemodynamic effects of hydrocortisone is consistent with those in previous studies. These hemodynamic effects may represent a beneficial role of hydrocortisone.”

Clinical decision support may be best strategy to reduce imaging for pulmonary embolism

Of the interventions tested to reduce the overuse of CT pulmonary angiography for pulmonary embolism, clinical decision support may have an edge over educational or hospital-policy interventions, according to a systematic review.

Researchers reviewed 17 studies, 15 that evaluated interventions in the ED and two that were conducted on clinical wards. The studies tested four types of intervention: electronic clinical decision support (eight studies), educational interventions (seven studies), performance feedback reports (one study), and an institutional clinical pretest policy (one study). Results were published in the January Journal of Hospital Medicine.

Electronic clinical decision support was associated with a reduction in the use of imaging ranging between 8.3% and 25.4%. It was also associated with a rise in diagnostic yield of 3.4% to 4.4% and an increase in appropriate ordering of 18% to 19%. In one study that evaluated clinical decision support combined with performance and feedback reports, appropriate ordering increased by 8.8%.

Educational intervention strategies involved training sessions to bolster physician use of clinical decision rules to diagnose pulmonary embolism. Postintervention, three studies reported a significant impact on adherence to clinical guidelines, two found a significant decrease in imaging use, and one found an increase in diagnostic yield.

One study that assessed the impact of hospital policy found a significant reduction in CT pulmonary angiography use and a significant increase in diagnostic yield. However, only 4% of patient charts reported a clinical probability of pulmonary embolism.

Limitations of the review include that the studies were heterogeneous and often did not evaluate the safety of the interventions. In addition, only three studies were experimental in design with a comparative control group.

The growing use of electronic health records implies that clinical decision support tools will be implemented across the spectrum of diagnoses, including pulmonary embolism, an accompanying editorial noted. “Moving forward, [clinical decision support], perhaps coupled with performance feedback, seems to offer the best hope of reducing inappropriate CT [pulmonary angiography] use,” the editorialists wrote.

No adverse events from MRI in patients with pacemakers and ICDs, study finds

More than a thousand patients with either a pacemaker or implantable cardioverter-defibrillator (ICD) underwent MRI without any long-term adverse effects, according to a study.

The prospective, nonrandomized study included 1,509 patients who had a “legacy” pacemaker (58%) or ICD (42%) that did not meet FDA criteria to be MRI-conditional. They underwent 2,103 thoracic and nonthoracic MRI examinations at magnetic field strength of 1.5 Tesla. The pacing mode was changed to asynchronous for pacing-dependent patients and to demand for other patients, and tachyarrhythmia functions were disabled. Results were published in the Dec. 28, 2017, New England Journal of Medicine.

In nine of the MRI exams, the device reset to a backup mode (0.4%; 95% CI, 0.2% to 0.7%). In eight of these, the reset was transient. One pacemaker, which had less than a month of battery life left, reset to ventricular inhibited pacing, could not be reprogrammed, and was subsequently replaced. The most common notable change in device parameters immediately after MRI was a decrease in P-wave amplitude; a decrease greater than 50% occurred in 1% of patients. No clinically significant adverse events were reported in long-term follow-up, during which the most common notable changes were decreases in P-wave amplitude (4%), increases in atrial capture threshold (4%), increases in right ventricular capture threshold (4%), and increases in left ventricular capture threshold (3%).

The results of this study, combined with a recent registry study, “provide complementary evidence that MRI scanning can be performed safely in patients with legacy devices, provided that an appropriate protocol is followed,” the authors said. They added numerous caveats, including “the need for appropriate device programming, monitoring by qualified personnel, and the availability of an external pacing backup” for pacing-dependent patients. Limitations of the study include that long-term follow-up information was not obtained from 20% of the patients and that all of the MRIs were performed at a field strength of 1.5 Tesla, so the results should not be extrapolated to higher or lower field strengths.

Machine learning model predicts sepsis in ICU patients up to 12 hours before diagnosis

A machine learning model accurately predicted the onset of sepsis in ICU patients four to 12 hours prior to clinical recognition in a study.

The observational cohort study used data from more than 31,000 admissions to ICUs at two academic hospitals as a development cohort and over 52,000 ICU patients from a publicly available database as a validation cohort. Patients who already met Sepsis-3 criteria were excluded. A set of 65 variables was calculated on an hourly basis and put into the Artificial Intelligence Sepsis Expert algorithm to predict the onset of sepsis within four, six, eight, or 12 hours. The algorithm also produced a list of the most significant contributing factors. Results were published by Critical Care Medicine on Dec. 26, 2017, and appeared in the April issue.

For all four time periods, the algorithm achieved an area under the receiver-operating characteristic curve (AUROC) in the range of 0.83 to 0.85, leading the study authors to conclude that it “can accurately predict the onset of sepsis in an ICU patient 4–12 hours prior to clinical recognition.” This was true whether the predicted outcome was objective physiologic manifestations of sepsis (the Sepsis-related Organ Failure Assessment score) or clinical suspicion of infection. The authors noted that the algorithm's predictive performance, as measured by the AUROC, specificity, and accuracy, decreased as the prediction window lengthened from four to 12 hours.

The next step for the tool will be to have its clinical utility analyzed in a prospective study, with the goal of having patients treated with IV antibiotics, fluids, and other adjunct therapies “based on a reliable estimate of their likelihood of developing sepsis in the near future,” the authors said. They plan to test this approach using a remote monitoring team, rather than sending alerts to bedside clinicians. The tool might also be useful in ICU patients who are decompensating due to causes other than sepsis, given that study patients who had high risk scores but did not develop sepsis (false positives) had twice the mortality risk of those who had low risk scores but developed sepsis (false negatives), the authors said.