Unplanned readmission to hospital is a critical quality improvement (QI) metric that can reflect the appropriateness and completeness of inpatient care. It impacts insurance reimbursements to health care facilities because of this, thereby providing a financial incentive for minimization. It is also a marker of health care disparities, as social determinants of health are known to influence hospital readmission risk. Despite the importance of this QI metric, it is challenging to accurately quantify. For example, readmissions to a nonindex hospital are difficult to capture, especially in larger cities with multiple health care centers. Furthermore, although readmissions <30 days is the standard timeline captured, this may undercount true readmission rates. Unplanned readmissions after Emergency General Surgery (EGS) diagnoses are especially important to benchmark because emergency surgery carries a heightened risk of complications and many EGS patients are un- or under-insured, known risk factors for readmission.
For these reasons, Urrechagaet al. endeavored to define the national rates of and risk factors for EGS readmissions in their recent article published in theJournal of Trauma and Acute Care Surgery. The authors also aimed to delineate the frequency of readmission to nonindex hospitals and compare risk factors for <30-day versus <90-day readmissions. The study utilized the Nationwide Readmissions Database (NRD) (2013-2014), which is unique in its inclusion of statewide readmissions to nonindex hospitals. Patients were included if they were admitted with one of 621 nonelective EGS diagnoses, as defined by the American Association for the Surgery of Trauma (AAST) based on ICD-9 codes. Exclusions were death, age <16 years, missing data, and readmissions not attributable to a single center. Study groups were defined as readmitted vs. non-readmitted patients within 30 days, with subgroup analyses of patients readmitted to nonindex hospitals and of readmissions <90 days.
After study criteria, 4,171,983 patients remained for analysis, of which 27% had undergone surgical intervention. Readmission <30 days occurred in 13%, of which 21% occurred to nonindex hospitals. When readmissions <90 days were examined, the percentage of readmitted patients grew to 22%, of which 23% occurred to nonindex hospitals. The most common indication for readmission to an index or nonindex hospital was infection. Readmission was most common among the uninsured and those with public insurance (Medicare and Medicaid) and least common among patients with private insurance. Nonindex hospital readmission occurred most frequently after discharge from for-profit hospitals and among the severely ill.
Multivariable logistic regression revealed the following risk factors for readmission <30 days: leaving against medical advice (odds ratio, OR 2.51); increased hospital length of stay, particularly >7 days (OR 2.04); Medicare insurance (OR 1.45); and Medicaid insurance (OR 1.38). Operative intervention was associated with reduced readmission (OR 0.61). Comparable risk factors were identified when the regression was repeated for readmission <90 days. Readmission to nonindex hospitals also carried similar risk factors, with the addition of being uninsured (OR 1.22). Index admission to a metropolitan hospital, both for nonteaching and teaching centers, was protective against nonindex hospital readmission (OR 0.85 and 0R 0.64). These risk factors were consistent when readmission was defined as <30 days versus <90 days. Study limitations include the database nature of the study, which hindered granularity in data capture (particularly in terms of patient race and ethnicity) and increased likelihood of coding errors. Out-of-state readmissions could not be captured and therefore actual readmissions may be underestimated by this study.
The most striking and actionable findings from this study include the fact that using <30 days as the cutoff point to define readmission underestimates the true burden of this metric, suggesting that utilization of <90 days may be a more appropriate benchmark moving forward. Next, the impact of insurance status and hospital type on readmissions is concerning, particularly as it exposes the possibility of health care disparities. Admission to a for-profit hospital was strongly associated with readmission. Particularly because readmissions reduce hospital reimbursements, this finding underscores the importance of capturing and accurately attributing nonindex hospital readmissions. Additionally, uninsured patients and those with public insurance were more likely to be readmitted, which is likely reflective of social determinants of health. As the study authors conclude, these identified disparities must be further studied so that the effects of insurance status and hospital type on readmission may be better understood and addressed.
Studies estimate that > 40% of EGS operations occur in adults whose age is > 65 and this percentage is expected to increase. An increasing proportion of these patients will come from nursing homes. This study aims to evaluate post – operative survival of older patients who come from nursing homes, as they are perceived to be a high-risk population due to presumed poor baseline functional status.
This is a retrospective cohort study of all older adults residing in nursing homes in Ontario, Canada who underwent hospital admission for an EGS (emergency general surgery) diagnosis between 04/2006 and 03/2018. The included EGS diagnoses were appendicitis, cholecystitis, diverticulitis, hernia with obstruction or strangulation, bowel obstruction, peptic ulcer disease, intestinal ischemia, or perforated viscus. They used an administrative dataset for this analysis. Patients were characterized as receiving was non- operative care versus surgical intervention. The primary outcome of interest was time spent alive in the year following EGS admission. Secondary outcomes hospital length of stay and in- hospital mortality.
A total of 7942 nursing home residents were admitted for an EGS condition and matched 1:1 with controls. The most common diagnoses were bowel obstruction, diverticulitis, and cholecystitis. Among those who underwent surgery, the most common surgical procedures were for bowel obstruction, hernia repair, and cholecystectomy. Median hospital LOS was 7 days. An ICU stay was required for 14% of the patients, with a median ICU LOS of 4 days. After 1 year of follow- up, the probability of survival was 55.0% for all EGS admissions and 72.1% for matched controls. Survival decreased as diagnosis severity increased. In patients who underwent surgery during their admission, the 1-year survival probability was 60.6% compared with 72.4% for matched controls. In patients managed non - operatively, the 1-year survival probability was 53.1% compared with 71.8% for matched controls. One year survival was higher in patients who did not require mechanical ventilation. Patients who required > 3 days of mechanical ventilation had a median hospital LOS of 17 days, and a 1-year survival probability ranging from 21.3% - 37.8% depending on EGS diagnosis severity.
This paper contributes to the literature evaluating survival of nursing home residents after emergency general surgery procedures. This information is important when discussing with patient and families as they weigh risks/benefits of operative intervention. The authors conclude that based on this information there is futility in operating on nursing home bound patients. They further conclude that aggressive medical and surgical care should not be withheld in this specific patient population 2/2 their living situation.
The purpose of this study was to evaluate the relative differences in in-hospital mortality when hypotensive trauma patients are transported by ambulance services staffed by physicians as opposed to those staffed by emergency medical services (EMS) staff alone.
The authors begin by discussing the existing literature regarding physician-staffed ambulance services, which despite relative rarity in the United States exist more broadly in various counties worldwide. These services are most commonly structured with one to two physicians working in conjunction with a crew of EMS personnel, and literature has reported relative benefits in specific settings such as airway management, traumatic brain injury, and out of hospital cardiac arrest. However, the effect of physician-staffed ambulance services has been met more controversially in the setting of the traumatically injured patient, with numerous studies reporting at best no survival benefit and at worst a longer duration to arrival at a setting for definitive care. The authors set out to better investigate the role of physician-staffed ambulance services in the care of the traumatically injured patient, both in regard to pre-hospital care as well as overall mortality.
This study utilized the Japan Trauma Data Bank (JTDB), the nationwide trauma data registry employed in Japan since its creation in 2003. The dataset was retrospectively reviewed for all trauma patients found to be hypotensive on scene (Systolic blood pressure <90 mmH) and transported directly to a trauma center, either by physician-staffed ambulance or EMS-alone staffed ambulance, over a roughly 16-year period (January 2004 - March 2019). The primary outcome measure of this study was in-hospital mortality, with secondary measures of cardiac arrest or hypotension on hospital arrival, length of stay, scene time interval, transport time, and total prehospital time. Importantly, groups were compared both on raw unadjusted data as well as following inverse probability weighting (IPW) to control for baseline characteristics such as age, comorbidity, mechanism of injury, on-scene vital signs, injury severity, and dispatch time.
Just over 14,000 patients were found to be eligible for this study, with 738 (5%) transported by physician-staffed ambulance services. When compared to EMS only transport, the physician-staffed transport showed increased in-hospital mortality and longer prehospital time both in raw and inverse probability weighted analysis. Physician-staffed transport had a 20% in-hospital mortality after IPW, as compared to 17% for EMS-only transport (OR 1.22; 95% confidence interval = 1.14-1.30). However, in subgroup analysis the inferior outcomes seen in physician-staffed transport was only observed in patients who arrived to the hospital with persistent hypotension (systolic pressure <90 mmHg).
The paper goes on to discuss potential reasons for the observed differences in mortality in hypotensive trauma patients, with the central idea that for traumatically injured patients manifesting hemodynamic instability, the critical interventions necessary for rescue are focused on surgical control of bleeding, and the increased transport time seen with physician-staffed transport can be assumed (but not proven) to translate to a relatively prolonged time to surgical hemorrhage control. The paper goes on to discuss that physician-staffed transport still has been shown in other studies to be associated with improved outcomes in the areas of pre-hospital cardiac arrest and non-surgical pathology, but in the case of trauma patients with hypotension, any delays to the delivery of definitive in-hospital surgical care can be hypothesized to translate to inferior outcomes. Admittedly this study is retrospective in nature and due to its use of large-population datasets lacks the desired granularity regarding specific interventions performed on each patient, but these are limitations of most to all retrospective large-volume datasets, and not this specific study in isolation. This paper provides a sound argument that transport time to definitive care is as important (or potentially more important) than the medical capabilities of transport mechanism in respect to the hypotensive trauma patient.
This Literature Review is being brought to you by the EAST Manuscript and Literature Review Committee. Have a suggestion for a review or an additional comment on articles reviewed? Please email litreview@east.org. Previous issues available on the EAST website.
