The Impact of a Machine Learning Early Warning Score on Hospital Mortality: A Multicenter Clinical Intervention Trial.

OBJECTIVES: To determine the impact of a machine learning early warning risk score, electronic Cardiac Arrest Risk Triage (eCART), on mortality for elevated-risk adult inpatients. DESIGN: A pragmatic pre- and post-intervention study conducted over the same 10-month period in 2 consecutive years. SETTING: Four-hospital community-academic health system. PATIENTS: All adult patients admitted to a medical-surgical ward. INTERVENTIONS: During the baseline period, clinicians were blinded to eCART scores. During the intervention period, scores were presented to providers. Scores greater than or equal to 95th percentile were designated high risk prompting a physician assessment for ICU admission. Scores between the 89th and 95th percentiles were designated intermediate risk, triggering a nurse-directed workflow that included measuring vital signs every 2 hours and contacting a physician to review the treatment plan. MEASUREMENTS AND MAIN RESULTS: The primary outcome was all-cause inhospital mortality. Secondary measures included vital sign assessment within 2 hours, ICU transfer rate, and time to ICU transfer. A total of 60,261 patients were admitted during the study period, of which 6,681 (11.1%) met inclusion criteria (baseline period n = 3,191, intervention period n = 3,490). The intervention period was associated with a significant decrease in hospital mortality for the main cohort (8.8% vs 13.9%; p < 0.0001; adjusted odds ratio [OR], 0.60 [95% CI, 0.52-0.71]). A significant decrease in mortality was also seen for the average-risk cohort not subject to the intervention (0.49% vs 0.26%; p < 0.05; adjusted OR, 0.53 [95% CI, 0.41-0.74]). In subgroup analysis, the benefit was seen in both high- (17.9% vs 23.9%; p = 0.001) and intermediate-risk (2.0% vs 4.0 %; p = 0.005) patients. The intervention period was also associated with a significant increase in ICU transfers, decrease in time to ICU transfer, and increase in vital sign reassessment within 2 hours. CONCLUSIONS: Implementation of a machine learning early warning score-driven protocol was associated with reduced inhospital mortality, likely driven by earlier and more frequent ICU transfer.

Making inpatient medication reconciliation patient centered, clinically relevant, and implementable: a consensus statement on key principles and necessary first steps.

This white paper identifies potential solutions to help ensure the utility and sustainability of this critical patient safety issue.

COVID-19 Vaccination and Parent-Reported Symptomatic Child Asthma Prevalence.

This cross-sectional study evaluates the association between COVID-19 vaccination and symptomatic child asthma.

Increases in Medicare prescription drug plan costs attributable to psychotropic medications.

OBJECTIVE: Older patients may regard some medications, particularly psychotropic medications, as discretionary compared with what they perceive as more "essential " nonpsychiatric medications. Patients' concerns about psychotropic medication costs under Medicare Part D may reinforce these impressions. DESIGN: The authors examined which Medicare prescription drug plans (PDPs) would be least expensive for beneficiaries considering the costs of 1) all medications; and 2) only nonpsychiatric medications. SETTING: The authors collected data from the PDP online comparison tool provided by the Centers for Medicare and Medicaid Services (CMS). PARTICIPANTS: Hypothetical Medicare beneficiaries. MEASUREMENTS: The authors examined four clinical scenarios from older outpatients with both chronic medical and psychiatric conditions (including psychosis, bipolar disorder, depression, and dementia with behavioral disturbances). RESULTS: The authors examined data from all 160 plans available in CMS PDP regions in May 2007. There were frequent discrepancies in the least expensive PDPs within region, depending on considering the costs of all medications, or just nonpsychiatric medications. In the clinical scenarios, patients selecting a PDP based on nonpsychiatric medications alone would pick an unnecessarily more expensive plan 74%-100% of the time (when they took any brand name medication), suggesting that excluding psychiatric medications from PDP choices may be excessively costly. However, brand name psychotropic medications significantly increased the costs of the least expensive plans. The latter finding might persuade patients to avoid taking needed psychiatric medication due to cost. CONCLUSION: This research highlights the complexity that patients with psychiatric and cognitive disorders face when choosing a Medicare PDP. Policymakers and clinicians should be aware of the tradeoffs that beneficiaries with psychiatric disorders face when making PDP plan choices.

Organizational structure for addressing the attributes of the ideal healthcare delivery system.

The Institute of Medicine's (IOM) report Crossing the Quality Chasm described the aims, characteristics, and components of the ideal healthcare system but did not provide the templates of organizational structures needed to achieve this vision. In this article, we review three principles of effective organizations to inform the design of a facilitative clinical care structure: a focus on the patient and caregiving team, the use of information, and connectivity with executive and operational leadership. These concepts can be realized in an organizational chart that is inverted to place patients and their care providers on top, flat with few degrees of separation between patients and executive leadership, and webbed to reflect connections to the professional and ancillary departments. An example of a recently implemented clinical care infrastructure follows this discussion. This model divides the patient population into nonexclusive subgroups, each with an interdisciplinary collaborative practice team that oversees and advocates the subgroup's clinical care activities. The organization's interdisciplinary practice council, in conjunction with its physician and nursing practice councils, backs these teams, providing a second layer of support. The council layer is connected to the health system board through the clinical oversight group, whose core membership consists of council chairs, the chief executive officer, and the chief medical and nursing officers. Clinical information for planning and evaluation is available at all levels. This model provides a framework for identifying the individuals and processes necessary to achieve IOM's vision.

Variation in estimated Medicare prescription drug plan costs and affordability for beneficiaries living in different states.

BACKGROUND: Medicare Part D prescription drug plans (PDPs) implemented in January 2006 are designed to improve beneficiaries' access to pharmaceuticals and use market competition to yield affordable drug costs. Variations in estimated PDP costs for beneficiaries living in different states have not previously been characterized. OBJECTIVE: To describe variations in the estimated costs of PDPs (plan premium, copays, and coinsurance) within and across states. DESIGN: To estimate PDP costs based on 4 actual patient cases that exemplify common conditions and prescription drug combinations for Medicare beneficiaries, we used the online tool provided by the Centers for Medicare and Medicaid Services. MEASUREMENTS: Principal study outcomes included (a) variation across states in the estimated annual cost of the lowest-cost PDP for each case and (b) variation in the estimated affordability of the lowest-cost PDPs across states, based on cost-of-living-adjusted median income for zero-earner households. RESULTS: For all 4 patient cases, we found substantive within-state and between-state differences in the estimated costs of Medicare PDPs incurred by beneficiaries. The estimated annual costs to beneficiaries of the lowest-cost PDPs varied across states by as much as $320 for medications in the least expensive scenario, and by as much as $13,000 for the most expensive scenario. On average across states, a beneficiary with cost-of-living-adjusted median income would expect to spend 3%-28% of annual income to pay for medications in the lowest-cost PDPs in the 4 patient cases. The affordability of the lowest-cost plans varied across states, and for 2 of the 4 cases the lowest-cost PDP estimates were negatively correlated with cost-of-living-adjusted median income. CONCLUSIONS: Substantive differences in estimated PDP costs are evident across states for patients with common Medicare conditions. Importantly, the lowest-cost plans were not proportionally affordable with respect to state-specific cost-of-living-adjusted median income. Refinement of the Medicare drug program may be needed to improve national balance in PDP affordability for beneficiaries living in different states.

A comparison of two hospitalist models with traditional care in a community teaching hospital.

PURPOSE: Many studies have documented significant length of stay reduction and cost savings when hospitalist care is compared with traditional care. However, less is known about the concurrent performance of more than one hospitalist model in a single site. SUBJECTS AND METHODS: This retrospective cohort study of 10595 patients was conducted between July 2001 and June 2002 in a tertiary care community-based teaching hospital. Risk-adjusted length of stay, variable costs, 30-day readmission rates, and in-hospital and 30-day mortality were measured for patients treated by Community Physicians, Private Hospitalists and Academic Hospitalists. RESULTS: There was a 20% reduction in length of stay on the Academic Hospitalist service (p <.0001) and 8% on the Private Hospitalist service (P = .049) compared with Community Physicians. Similarly, total costs were 10% less on the Academic (P <.0001) and 6% less on the Private Hospitalist (P = .02) services compared with Community Physicians. The length of stay of Academic Hospitalists was 13% shorter than that of Private Hospitalists (P = .002); differences in costs between hospitalist groups were not statistically significant. Differences in in-hospital and 30-day mortality and 30-day readmission rates among the 3 physician groups were also not statistically significant. CONCLUSIONS: The impact on patient outcomes and resource utilization may vary with the hospitalist model used. Future studies should examine the specific organizational characteristics of hospitalists that contribute to improved patient care and resource utilization.

Provider response to different formats of the adult immunization schedule.

BACKGROUND: Providers' failure to administer vaccines in accordance with established recommendations is a well-recognized barrier to national immunization efforts. This study evaluated the ease of use of two different formats of the Centers for Disease Control and Prevention's (CDC) adult immunization schedule by physicians in private practice, where the majority of adult immunizations are administered. METHODS: A series of focus groups was conducted with 94 physicians and other clinical staff in 11 private practices (family medicine and internal medicine) in six U.S. cities. Each session was based on a structured set of questions that explored barriers to adult immunizations, followed by three mock clinical scenarios to examine how each of two graphical depictions of the 2003-2004 adult immunization schedule (one from the CDC's Advisory Committee on Immunization Practices, and the other from the Immunization Action Coalition) might facilitate assessments of recommended immunizations. Group dialogue and individual participants' written responses to the scenarios and the alternate schedule formats were analyzed. RESULTS: Providers perceived multiple barriers to adult immunization independent of immunization schedule formats, chiefly patients' low interest in immunization and refusal of vaccines. Most participants were not familiar with either format of CDC's adult immunization schedule before the study, but quickly developed strong preferences for one versus the other (usually the second format that they encountered). About half of the providers changed their vaccine recommendations for clinical scenarios when they consulted either schedule format, although some of the changes were not clinically appropriate. Participants suggested several ways to enhance the availability of the information contained in the schedule formats, especially through electronic means. CONCLUSIONS: This qualitative study suggests ways in which graphic depictions of an adult immunization schedule may address adult immunization barriers. Greater provider familiarity with schedule formats will be critical to their appropriate application in clinical encounters.

Implementation of a mortality prediction rule for real-time decision making: feasibility and validity.

BACKGROUND: A previously published, retrospectively derived prediction rule for death within 30 days of hospital admission has the potential to launch parallel interdisciplinary team activities. Whether or not patient care improves will depend on the validity of prospectively generated predictions, and the feasibility of generating them on demand for a critical proportion of inpatients. OBJECTIVE: To determine the feasibility of generating mortality predictions on admission and to validate their accuracy using the scoring weights of the retrospective rule. DESIGN: Prospective, sequential cohort. SETTING: Large, tertiary care, community hospital in the Midwestern United States PATIENTS: Adult patients admitted from the emergency department or scheduled for elective surgery RESULTS: Mortality predictions were generated on demand at the beginning of the hospitalization for 9312 (92.9%) out of a possible 10,027 cases. The area under the receiver operating curve for 30-day mortality was 0.850 (95% confidence interval: 0.833-0.866), indicating very good to excellent discrimination. The prospectively generated 30-day mortality risk had a strong association with the receipt of palliative care by hospital discharge, in-hospital mortality, and 180-day mortality, a fair association with the risk for 30-day readmissions and unplanned transfers to intensive care, and weak associations with receipt of intensive unit care ever within the hospitalization or the development of a new diagnosis that was not present on admission (ie, complication). CONCLUSIONS: Important prognostic information is feasible to obtain in a real-time, single-assessment process for a sizeable proportion of hospitalized patients.

Spinal and paraspinal fungal infections associated with contaminated methylprednisolone injections.

BACKGROUND: A nationwide outbreak of fungal infections was traced to injection of Exserohilum-contaminated methylprednisolone. We describe our experience with patients who developed spinal or paraspinal infection after injection of contaminated methylprednisolone. METHODS: Data were assembled from the Michigan Department of Community Health, electronic medical records, and magnetic resonance imaging (MRI) reports. RESULTS: Of 544 patients who received an epidural injection from a contaminated lot of methylprednisolone at a pain clinic in southeastern Michigan, 153 (28%) were diagnosed at our institution with probable or confirmed spinal or paraspinal fungal infection at the injection site. Forty-one patients had both meningitis and spinal or paraspinal infection, and 112 had only spinal or paraspinal infection. Magnetic resonance imaging abnormalities included abscess, phlegmon, arachnoiditis, and osteomyelitis. Surgical debridement in 116 patients revealed epidural phlegmon and epidural abscess most often. Among 26 patients with an abnormal MRI but with no increase or change in chronic pain, 19 (73%) had infection identified at surgery. Fungal infection was confirmed in 78 patients (51%) by finding hyphae in tissues, positive polymerase chain reaction, or culture. Initial therapy was voriconazole plus liposomal amphotericin B in 115 patients (75%) and voriconazole alone in 38 patients (25%). As of January 31, 2014, 20 patients remained on an azole agent. Five patients died of infection. CONCLUSIONS: We report on 153 patients who had spinal or paraspinal fungal infection at the site of epidural injection of contaminated methylprednisolone. One hundred sixteen (76%) underwent operative debridement in addition to treatment with antifungal agents.

Mortality predictions on admission as a context for organizing care activities.

BACKGROUND: Favorable health outcomes are more likely to occur when the clinical team recognizes patients at risk and intervenes in consort. Prediction rules can identify high-risk subsets, but the availability of multiple rules for various conditions present implementation and assimilation challenges. METHODS: A prediction rule for 30-day mortality at the beginning of the hospitalization was derived in a retrospective cohort of adult inpatients from a community hospital in the Midwestern United States from 2008 to 2009, using clinical laboratory values, past medical history, and diagnoses present on admission. It was validated using 2010 data from the same and from a different hospital. The calculated mortality risk was then used to predict unplanned transfers to intensive care units, resuscitation attempts for cardiopulmonary arrests, a condition not present on admission (complications), intensive care unit utilization, palliative care status, in-hospital death, rehospitalizations within 30 days, and 180-day mortality. RESULTS: The predictions of 30-day mortality for the derivation and validation datasets had areas under the receiver operating characteristic curve of 0.88. The 30-day mortality risk was in turn a strong predictor for in-hospital death, palliative care status, 180-day mortality; a modest predictor for unplanned transfers and cardiopulmonary arrests; and a weaker predictor for the other events of interest. CONCLUSIONS: The probability of 30-day mortality provides health systems with an array of prognostic information that may provide a common reference point for organizing the clinical activities of the many health professionals involved in the care of the patient.

Hospitals during recession and recovery: vulnerable institutions and quality at risk.

For generations, American hospitals have been considered recession-proof, but there is reason to believe the current economic crisis is an exception. Hospitals have shown declining financial margins and decreased admissions. The severe recession has adversely affected many hospitals' finances, creating a risk of closure and constraining plans for expansion. We believe there is also a risk of harming clinical quality, through decreased staffing that may limit the momentum of the hospital quality movement, especially in fiscally vulnerable institutions. We consider ways the federal government could aid hospitals by promoting hospital quality while providing employment.

Conflicting measures of hospital quality: ratings from "Hospital Compare" versus "Best Hospitals".

BACKGROUND: In April 2005 the Centers for Medicare and Medicaid Services launched "Hospital Compare," the first government-sponsored hospital quality scorecard. We compared the ranking of U.S. News and World Report's "Best Hospitals" with Hospital Compare performance ratings. METHODS: We examined Hospital Compare scores for core measures related to care for acute myocardial infarction (AMI), congestive heart failure (CHF), and community-acquired pneumonia (CAP). We calculated composite scores for the disease-specific sets of core measures and a composite combined score for the 14 core measures (across 3 diseases) and determined national score quartile cut points for each set. We then characterized the quartile distribution of Hospital Compare scores for the Best Hospitals for care of cardiac conditions and respiratory disorders in each year, as well as for the Best Hospital "Honor Roll" institutions. RESULTS: AMI scores were available for 2165 hospitals, CHF scores for 3130, and CAP scores for 3462. In both 2004 and 2005, fewer than 50% of the Best Hospitals for cardiac care rated in the top quartile of Hospital Compare scores for AMI and CHF. Among the Best Hospitals for care of respiratory disorders, fewer than 15% scored in the top Hospital Compare quartile for CAP. Among Honor Roll institutions, only 5 (of 14 hospitals in 2004; of 16 in 2005) ranked in the top quartile for the combined core measure score. CONCLUSIONS: Hospital Compare scores are frequently discordant with Best Hospital rankings, which is likely attributable to the markedly different methods each rating approach employs. Such discordance between major quality rating systems paints a conflicting picture of institutional performance for the public to interpret.

An electronic strategy to identify hospitalized heart failure patients.

BACKGROUND: A common challenge in improving performance measures regarding heart failure (HF) is identifying patients early in the course of their hospitalization so that multidisciplinary education and clinical interventions can be implemented. We describe the accuracy of using an electronic pharmacy-based strategy to identify hospitalized patients likely to have a principal diagnosis of HF at discharge. METHODS: We evaluated 2 strategies. The first used the receipt of an intravenous loop diuretic as a single predictor; the second incorporated additional lab, pharmacy, and demographic information in a multivariable general estimating equation. RESULTS: Receipt of an intravenous diuretic predicted a discharge diagnosis of heart failure with a sensitivity of 0.89 and a specificity of 0.87. Adding age, B-type natriuretic peptide level, previous hospitalizations, attending physician specialty, and receipt of spironolactone into the predictor improved the sensitivity to 0.91 and the specificity to 0.89. CONCLUSIONS: The receipt of intravenous loop diuretics is a reasonable and easily implemented screening test to identify patients likely to have a principal diagnosis of heart failure at discharge. The accuracy is improved by incorporating other electronically available variables.

Predicting life expectancy in men with clinically localized prostate cancer.

PURPOSE: The choice of appropriate management for localized prostate cancer depends in part on the estimated life expectancy of a man. Little is known about the accuracy of existing rules for deriving these estimates. We developed a new prediction rule and examined the accuracy of 2 others in our data set. MATERIALS AND METHODS: A retrospective cohort was assembled, consisting of 506 men who were diagnosed or received initial treatment at a community based, tertiary care health center between 1987 and 1989 for clinically localized prostate cancer (stages A, B, I, II or T1-2N0M0) and had at least 13 years of followup. Most patients did not have prostate specific antigen levels available. Proportional hazards regression was used to create a nomogram for deriving survival estimates. Discrimination of the new and external prediction rules was assessed by the c-statistic. Calibration curves compared predicted to actual survival at 10 years. RESULTS: Estimates for survival at 5, 10 and 15 years, and for median life expectancy were determined. Discrimination was modest with a c-statistic of 0.73. The rules of Albertsen and Tewari et al had comparable discrimination in our data with a c-statistic of 0.71 and 0.70, respectively. Predicted life expectancy according to our rule and that of Tewari approximated actual survival experience. Predictions according to the Albertsen study underestimated actual survival in our group but in consistent fashion. CONCLUSIONS: Overall life expectancy can be predicted with a moderate degree of accuracy, sufficient for informing patient-clinician discussions but inadequate as the only determinant of the optimal management approach.