SAN DIEGO — The results of two highlighted studies presented here demonstrated the merit of using electronic health record (EHR) data to improve patient monitoring.
Filling Surveillance Gaps in General Care
A new risk prediction tool could prevent respiratory compromise and decrease unplanned intensive care unit (ICU) admissions, preliminary results from the PRODIGY trial found.
The model uses age, sex, opioid naivety, and the presence of sleep disorders or hypertension to determine what researchers call a PRODIGY score, with higher scores indicating greater risk for opioid-induced respiratory depression, according to Ashish Khanna, MD, of Wake Forest School of Medicine in Winston-Salem, North Carolina.
PRODIGY scores successfully predicted 76% of patients with confirmed respiratory depression (area under the curve 0.7676), and had low optimism between the derivation and validation models (0.001), Khanna reported here at the Society of Critical Care Medicine (SCCM) meeting.
“The issue is that we send our patients to the hospital general care floors, hospital wards, or nursing floors and normally we assume patients sent to these areas are relatively healthy and that they are going to transition out of the hospital and likely go home,” he told MedPage Today. “But the unfortunate reality is that far too often these patients who are recovering on hospital general care floors are faced with sudden unprecedented cardiovascular decompensation events.”
Khanna noted this model detected 46% of patients that experienced opioid-induced respiratory depression, a percentage up to twice as high as some prior estimates.
The researchers used data from a prior PRODIGY study in which Capnostream respiratory monitors measured continuous capnography and oximetry in a group of patients at 16 sites across the U.S., Europe, and Asia. Screens were concealed and alarms were silenced throughout a 24-hour monitoring period to ensure blinding, although spot-check monitoring continued unblinded, they reported.
Respiratory depression events were defined as an end-tidal carbon dioxide ≤15 or ≥60 mm Hg for at least 3 minutes, a respiratory rate of ≤5 breaths for at least 3 minutes, oxygen saturation of ≤85% for for at least 3 minutes, apnea for more than 30 seconds, or a respiratory opioid-related adverse event.
Oversight for Hospital-Onset Sepsis
Clinical data from over 100 U.S. hospitals may suggest lapses in quality of care for patients with hospital-onset sepsis as well, according to research presented here.
In a sample of nearly 100,000 patients with sepsis, 1 in 8 cases were hospital-onset and these patients were twice as likely to die compared to patients with community-onset sepsis (OR 2.10, 95% CI 2.08–2.12), and three times more likely to die than patients without sepsis (HR 3.02, 95% CI 2.99–3.04) after adjusting for demographics, comorbidities, infectious diagnoses, and severity-of-illness at sepsis onset or on admission, reported Chanu Rhee, MD, MPH, of Harvard Medical School in Boston.
“The higher risk of mortality in hospital versus community-onset sepsis suggests potential differences in quality of care that explain the higher mortality,” Rhee said. “Higher mortalities associated with hospital-onset sepsis argue that some or maybe even many of these deaths may be preventable.”
Compared to those who had community-onset sepsis, individuals who developed hospital-onset sepsis more often had heart failure (26% vs 22%), renal disease (23% vs 20%), and cancer (17% vs 11%), Rhee reported.
In the study, patients with hospital-onset sepsis also had significantly higher mean Sequential Organ Failure Assessment (SOFA) scores than community-onset sepsis patients (4 vs 3), higher rates of intra-abdominal infections (20% vs 15%), more positive blood cultures (26% vs 21%), longer median hospital length-of-stays (19 vs 8 days), and median ICU length-of-stays (6 vs 4 days). They were also more likely to die in the hospital (34% vs 17%).
Researchers applied the CDC Adult Sepsis Event criteria to EHR data from 136 U.S. hospitals in the Cerner Health Facts dataset from patients hospitalized from 2009 to 2015. Hospital-onset sepsis was defined by blood cultures, antibiotics, and organ dysfunction on hospital day 3 or later, while community-onset sepsis was classified by blood cultures, antibiotics, or organ dysfunction prior to hospital day 3.
In total, 97,352 patients had sepsis and 11,782 of those cases were hospital-onset (12.1%). Patients with community-onset sepsis tended to be older than those with hospital-onset sepsis (68 vs 66 years).
Limitations of the Studies
Khanna said that his study is limited because it has not been independently validated outside the study. He also noted that it’s possible methodological variations occurred across the sites.
Rhee said admission codes are often misused and can be variably applied across hospitals, causing inaccuracies in the number of hospital-onset cases that are recorded. The study was also limited because the researchers used administrative data to identify the source of the infection and did not examine differences in care processes. Additionally, data on some covariates, such as patient insurance or admitting services, were not available, and some subtle markers of frailty or comorbidities could not easily be measured by EHR data, he said.
Khanna received consulting fees for the PRODIGY Trial Steering Committee and the Cleveland Clinic received grant funding from Medtronic for the study.
Rhee did not report any relevant disclosures.