The increasing impact of length of stay "outliers" on length of stay at an urban academic hospital.

BACKGROUND: As healthcare systems strive for efficiency, hospital "length of stay outliers" have the potential to significantly impact a hospital's overall utilization. There is a tendency to exclude such "outlier" stays in local quality improvement and data reporting due to their assumed rare occurrence and disproportionate ability to skew mean and other summary data. This study sought to assess the influence of length of stay (LOS) outliers on inpatient length of stay and hospital capacity over a 5-year period at a large urban academic medical center. METHODS: From January 2014 through December 2019, 169,645 consecutive inpatient cases were analyzed and assigned an expected LOS based on national academic center benchmarks. Cases in the top 1% of national sample LOS by diagnosis were flagged as length of stay outliers. RESULTS: From 2014 to 2019, mean outlier LOS increased (40.98 to 45.11 days), as did inpatient LOS with outliers excluded (5.63 to 6.19 days). Outlier cases increased both in number (from 297 to 412) and as a percent of total discharges (0.98 to 1.56%), and outlier patient days increased from 6.7 to 9.8% of total inpatient plus observation days over the study period. CONCLUSIONS: Outlier cases utilize a disproportionate and increasing share of hospital resources and available beds. The current tendency to exclude such outlier stays in data reporting due to assumed rare occurrence may need to be revisited. Outlier stays require distinct and targeted interventions to appropriately reduce length of stay to both improve patient care and maintain hospital capacity.

Cost-effectiveness of a Digital Health Intervention for Acute Myocardial Infarction Recovery.

BACKGROUND: Acute myocardial infarction (AMI) is a common cause of hospital admissions, readmissions, and mortality worldwide. Digital health interventions (DHIs) that promote self-management, adherence to guideline-directed therapy, and cardiovascular risk reduction may improve health outcomes in this population. The "Corrie" DHI consists of a smartphone application, smartwatch, and wireless blood pressure monitor to support medication tracking, education, vital signs monitoring, and care coordination. We aimed to assess the cost-effectiveness of this DHI plus standard of care in reducing 30-day readmissions among AMI patients in comparison to standard of care alone. METHODS: A Markov model was used to explore cost-effectiveness from the hospital perspective. The time horizon of the analysis was 1 year, with 30-day cycles, using inflation-adjusted cost data with no discount rate. Currencies were quantified in US dollars, and effectiveness was measured in quality-adjusted life-years (QALYs). The results were interpreted as an incremental cost-effectiveness ratio at a threshold of $100,000 per QALY. Univariate sensitivity and multivariate probabilistic sensitivity analyses tested model uncertainty. RESULTS: The DHI reduced costs and increased QALYs on average, dominating standard of care in 99.7% of simulations in the probabilistic analysis. Based on the assumption that the DHI costs $2750 per patient, use of the DHI leads to a cost-savings of $7274 per patient compared with standard of care alone. CONCLUSIONS: Our results demonstrate that this DHI is cost-saving through the reduction of risk for all-cause readmission following AMI. DHIs that promote improved adherence with guideline-based health care can reduce hospital readmissions and associated costs.

Digital Health Intervention in Acute Myocardial Infarction.

BACKGROUND: Thirty-day readmissions among patients with acute myocardial infarction (AMI) contribute to the US health care burden of preventable complications and costs. Digital health interventions (DHIs) may improve patient health care self-management and outcomes. We aimed to determine if patients with AMI using a DHI have lower 30-day unplanned all-cause readmissions than a historical control. METHODS: This nonrandomized controlled trial with a historical control, conducted at 4 US hospitals from 2015 to 2019, included 1064 patients with AMI (DHI n=200, control n=864). The DHI integrated a smartphone application, smartwatch, and blood pressure monitor to support guideline-directed care during hospitalization and through 30-days post-discharge via (1) medication reminders, (2) vital sign and activity tracking, (3) education, and (4) outpatient care coordination. The Patient Activation Measure assessed patient knowledge, skills, and confidence for health care self-management. All-cause 30-day readmissions were measured through administrative databases. Propensity score-adjusted Cox proportional hazard models estimated hazard ratios of readmission for the DHI group relative to the control group. RESULTS: Following propensity score adjustment, baseline characteristics were well-balanced between the DHI versus control patients (standardized differences <0.07), including a mean age of 59.3 versus 60.1 years, 30% versus 29% Women, 70% versus 70% White, 54% versus 54% with private insurance, 61% versus 60% patients with a non ST-elevation myocardial infarction, and 15% versus 15% with high comorbidity burden. DHI patients were predominantly in the highest levels of patient activation for health care self-management (mean score 71.7±16.6 at 30 days). The DHI group had fewer all-cause 30-day readmissions than the control group (6.5% versus 16.8%, respectively). Adjusting for hospital site and a propensity score inclusive of age, sex, race, AMI type, comorbidities, and 6 additional confounding factors, the DHI group had a 52% lower risk for all-cause 30-day readmissions (hazard ratio, 0.48 [95% CI, 0.26-0.88]). Similar results were obtained in a sensitivity analysis employing propensity matching. CONCLUSIONS: Our results suggest that in patients with AMI, the DHI may be associated with high patient activation for health care self-management and lower risk of all-cause unplanned 30-day readmissions. Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT03760796.

Relapse risk factors for patients with comorbid affective disorders and substance abuse disorders from an intensive treatment unit.

BACKGROUND AND OBJECTIVES: The prevalence of substance use disorders (SUD), particularly involving opiates and benzodiazepines, has increased to the detriment of public health and the economy. Here, we evaluate relapse factors among the high-risk demographic of patients with SUD and comorbid affective disorders. METHODS: A retrospective chart review of 76 patients discharged after detoxification and simultaneous psychiatric care for concomitant affective disorders and SUDs. Relapse was assessed by two independent evaluators via postdischarge chart review, which included state-wide healthcare utilization, by patient, through healthcare information exchange systems. A Cox Hazards analysis was performed to characterize relapse risk factors. RESULTS: Benzodiazepine use, admission through the emergency department (ED) rather than direct admission, frequent ED use in the preceding year, and history of prior attendance at multiple detoxification programs were risk factors for shortened time-to-relapse. Polysubstance use and intravenous drug use prolonged time to relapse. DISCUSSION AND CONCLUSIONS: Notable findings include the significant relapse risk associated with benzodiazepine abuse and frequent prior ED utilization. These risk factors could reflect a number of underlying mediators for relapse, including anxiety, disease burden, and malingering. Additionally, this study recapitulates the observation in other patient populations that the majority of health resource utilization is attributed to a small population of patients. SCIENTIFIC SIGNIFICANCE: This study is the first to identify relapse predictors among dual-diagnosis affective disorder and SUD patients in survival analysis, and replicates the alarming and largely unknown effect that benzodiazepines have on increasing relapse risk.

The Paradox of Readmission Prevention Interventions: Missing Those Most in Need.

BACKGROUND: Post-hospitalization transition interventions remain a priority in preventing rehospitalization. However, not all patients referred for readmission prevention interventions receive them. We sought to 1) define patient characteristics associated with non-receipt of readmission prevention interventions (among those eligible for them), and 2) determine whether these same patient characteristics are associated with hospital readmission at the state level. METHODS: We used state-wide data from the Maryland Health Services Cost Review Commission to determine patient-level factors associated with state-wide readmissions. Concurrently, we conducted a retrospective analysis of discharged patients referred to receive 1 of 3 post-discharge interventions between January 2013 and July 2019-a nurse transition guide, post-discharge phone call, or follow-up appointment in our post-discharge clinic-to determine patient-level factors associated with not receiving the intervention. Multivariable generalized estimating equation logistic regression models were used to calculate the odds of not accepting or not receiving the interventions. RESULTS: Older age, male gender, black race, higher expected readmission rate, and lower socioeconomic status were significantly associated with 30-day readmission in hospitalized Maryland patients. Most of these variables (age, sex, race, payer type [Medicaid or non-Medicaid], and socioeconomic status) were also associated with non-receipt of intervention. CONCLUSIONS: We found that many of the same patient-level characteristics associated with the highest readmission risk are also associated with non-receipt of readmission reduction interventions. This highlights the paradox that patients at high risk of readmission are least likely to accept or receive interventions for preventing readmission. Identifying strategies to engage hard-to-reach high-risk patients continues to be an unmet challenge in readmission prevention.

Association of a Care Coordination Model With Health Care Costs and Utilization: The Johns Hopkins Community Health Partnership (J-CHiP).

Importance: The Johns Hopkins Community Health Partnership was created to improve care coordination across the continuum in East Baltimore, Maryland. Objective: To determine whether the Johns Hopkins Community Health Partnership (J-CHiP) was associated with improved outcomes and lower spending. Design, Setting, and Participants: Nonrandomized acute care intervention (ACI) and community intervention (CI) Medicare and Medicaid participants were analyzed in a quality improvement study using difference-in-differences designs with propensity score-weighted and matched comparison groups. The study spanned 2012 to 2016 and took place in acute care hospitals, primary care clinics, skilled nursing facilities, and community-based organizations. The ACI analysis compared outcomes of participants in Medicare and Medicaid during their 90-day postacute episode with those of a propensity score-weighted preintervention group at Johns Hopkins Community Health Partnership hospitals and a concurrent comparison group drawn from similar Maryland hospitals. The CI analysis compared changes in outcomes of Medicare and Medicaid participants with those of a propensity score-matched comparison group of local residents. Interventions: The ACI bundle aimed to improve transition planning following discharge. The CI included enhanced care coordination and integrated behavioral support from local primary care sites in collaboration with community-based organizations. Main Outcomes and Measures: Utilization measures of hospital admissions, 30-day readmissions, and emergency department visits; quality of care measures of potentially avoidable hospitalizations, practitioner follow-up visits; and total cost of care (TCOC) for Medicare and Medicaid participants. Results: The CI group had 2154 Medicare beneficiaries (1320 [61.3%] female; mean age, 69.3 years) and 2532 Medicaid beneficiaries (1483 [67.3%] female; mean age, 55.1 years). For the CI group's Medicaid participants, aggregate TCOC reduction was $24.4 million, and reductions of hospitalizations, emergency department visits, 30-day readmissions, and avoidable hospitalizations were 33, 51, 36, and 7 per 1000 beneficiaries, respectively. The ACI group had 26 144 beneficiary-episodes for Medicare (13 726 [52.5%] female patients; mean patient age, 68.4 years) and 13 921 beneficiary-episodes for Medicaid (7392 [53.1%] female patients; mean patient age, 52.2 years). For the ACI group's Medicare participants, there was a significant reduction in aggregate TCOC of $29.2 million with increases in 90-day hospitalizations and 30-day readmissions of 11 and 14 per 1000 beneficiary-episodes, respectively, and reduction in practitioner follow-up visits of 41 and 29 per 1000 beneficiary-episodes for 7-day and 30-day visits, respectively. For the ACI group's Medicaid participants, there was a significant reduction in aggregate TCOC of $59.8 million and the 90-day emergency department visit rate decreased by 133 per 1000 episodes, but hospitalizations increased by 49 per 1000 episodes and practitioner follow-up visits decreased by 70 and 182 per 1000 episodes for 7-day and 30-day visits, respectively. In total, the CI and ACI were associated with $113.3 million in cost savings. Conclusions and Relevance: A care coordination model consisting of complementary bundled interventions in an urban academic environment was associated with lower spending and improved health outcomes.

Patterns of Hospital Performance on the Hospital-Wide 30-Day Readmission Metric: Is the Playing Field Level?

BACKGROUND: Hospital performance on the 30-day hospital-wide readmission (HWR) metric as calculated by the Centers for Medicare and Medicaid Services (CMS) is currently reported as a quality measure. Focusing on patient-level factors may provide an incomplete picture of readmission risk at the hospital level to explain variations in hospital readmission rates. OBJECTIVE: To evaluate and quantify hospital-level characteristics that track with hospital performance on the current HWR metric. DESIGN: Retrospective cohort study. SETTING/PATIENTS: A total of 4785 US hospitals. METRICS: We linked publically available data on individual hospitals published by CMS on patient-level adjusted 30-day HWR rates from July 1, 2011, through June 30, 2014, to the 2014 American Hospital Association annual survey. Primary outcome was performance in the worst CMS-calculated HWR quartile. Primary hospital-level exposure variables were defined as: size (total number of beds), safety net status (top quartile of disproportionate share), academic status [member of the Association of American Medical Colleges (AAMC)], National Cancer Institute Comprehensive Cancer Center (NCI-CCC) status, and hospital services offered (e.g., transplant, hospice, emergency department). Multilevel regression was used to evaluate the association between 30-day HWR and the hospital-level factors. RESULTS: Hospital-level characteristics significantly associated with performing in the worst CMS-calculated HWR quartile included: safety net status [adjusted odds ratio (aOR) 1.99, 95% confidence interval (95% CI) 1.61-2.45, p < 0.001], large size (> 400 beds, aOR 1.42, 95% CI 1.07-1.90, p = 0.016), AAMC alone status (aOR 1.95, 95% CI 1.35-2.83, p < 0.001), and AAMC plus NCI-CCC status (aOR 5.16, 95% CI 2.58-10.31, p < 0.001). Hospitals with more critical care beds (aOR 1.26, 95% CI 1.02-1.56, p = 0.033), those with transplant services (aOR 2.80, 95% CI 1.48-5.31,p = 0.001), and those with emergency room services (aOR 3.37, 95% CI 1.12-10.15, p = 0.031) demonstrated significantly worse HWR performance. Hospice service (aOR 0.64, 95% CI 0.50-0.82, p < 0.001) and having a higher proportion of total discharges being surgical cases (aOR 0.62, 95% CI 0.50-0.76, p < 0.001) were associated with better performance. LIMITATION: The study approach was not intended to be an alternate readmission metric to compete with the existing CMS metric, which would require a re-examination of patient-level data combined with hospital-level data. CONCLUSION: A number of hospital-level characteristics (such as academic tertiary care center status) were significantly associated with worse performance on the CMS-calculated HWR metric, which may have important health policy implications. Until the reasons for readmission variability can be addressed, reporting the current HWR metric as an indicator of hospital quality should be reevaluated.

Improving Outcomes After Hospitalization: A Prospective Observational Multicenter Evaluation of Care Coordination Strategies for Reducing 30-Day Readmissions to Maryland Hospitals.

BACKGROUND: Patients frequently experience suboptimal transitions from the hospital to the community, which can increase the likelihood of readmission. It is not known which care coordination services can lead to improvements in readmission rates. OBJECTIVE: To evaluate the effects of two care coordination interventions on 30-day readmission rates. DESIGN: Prospective multicenter observational study of hospitalized patients eligible for two care coordination services between January 1, 2013, and October 31, 2015. Readmission rates were compared for patients who received each care coordination intervention versus those who did not using multivariable generalized estimating equation logistic regression models. PARTICIPANTS: A total of 25,628 patients hospitalized in medicine, neurosciences, or surgical sciences units. INTERVENTIONS: Patients discharged home and deemed to be at high risk for readmission were assigned a nurse Transition Guide (TG) for 30 days post-discharge. All other patients were assigned the Patient Access Line (PAL) intervention, which provided a post-discharge phone call from a registered nurse. SETTING: Two large academic hospitals in Baltimore, MD. MAIN MEASURES: Thirty-day all-cause readmission to any Maryland hospital. KEY RESULTS: Among all patients, 14.2% (2409/16,993) of those referred for the PAL intervention and 22.8% (1973/8635) of those referred for the TG intervention were readmitted. PAL-referred patients who did not receive the intervention had an adjusted odds ratio (aOR) for readmission of 1.27 (95% confidence interval [95% CI] 1.12-1.44, p < 0.001) compared with patients who did. TG-referred patients who did not receive the TG intervention had an aOR of 1.83 (95% CI 1.60-2.10, p < 0.001) compared with patients who received the intervention. Younger age, male sex, having more comorbidities, and being discharged from a medicine unit were associated with not receiving an assigned intervention. These characteristics were also associated with higher readmission rates. CONCLUSIONS: PAL and TG care coordination interventions were associated with lower rates of 30-day readmission. Our findings underscore the importance of determining the appropriate intervention for the hardest-to-reach patients, who are also at the highest risk of being readmitted.

A Method for Attributing Patient-Level Metrics to Rotating Providers in an Inpatient Setting.

BACKGROUND: Individual provider performance drives group metrics, and increasingly, individual providers are held accountable for these metrics. However, appropriate attribution can be challenging, particularly when multiple providers care for a single patient. OBJECTIVE: We sought to develop and operationalize individual provider scorecards that fairly attribute patient-level metrics, such as length of stay and patient satisfaction, to individual hospitalists involved in each patient's care. DESIGN: Using patients cared for by hospitalists from July 2010 through June 2014, we linked billing data across each hospitalization to assign "ownership" of patient care based on the type, timing, and number of charges associated with each hospitalization (referred to as "provider day weighted "). These metrics were presented to providers via a dashboard that was updated quarterly with their performance (relative to their peers). For the purposes of this article, we compared the method we used to the traditional method of attribution, in which an entire hospitalization is attributed to 1 provider, based on the attending of record as labeled in the administrative data. RESULTS: Provider performance in the 2 methods was concordant 56% to 75% of the time for top half versus bottom half performance (which would be expected to occur by chance 50% of the time). While provider percentile differences between the 2 methods were modest for most providers, there were some providers for whom the methods yielded dramatically different results for 1 or more metrics. CONCLUSION: We found potentially meaningful discrepancies in how well providers scored (relative to their peers) based on the method used for attribution. We demonstrate that it is possible to generate meaningful provider-level metrics from administrative data by using billing data even when multiple providers care for 1 patient over the course of a hospitalization.

Unplanned 30-day hospital readmission as a quality measure in gynecologic oncology.

OBJECTIVES: Thirty-day readmission is used as a quality measure for patient care and Medicare-based hospital reimbursement. The primary study objective was to describe the 30-day readmission rate to an academic gynecologic oncology service. Secondary objectives were to identify risk factors and costs related to readmission. METHODS: This was a retrospective, concurrent cohort study of all surgical admissions to an academic, high volume gynecologic oncology service during a two-year period (2013-2014). Data were collected on patient demographics, medical comorbidities, psychosocial risk factors, and results from a hospital discharge screening survey. Mixed logistic regression was used to identify factors associated with 30-day readmission and costs of readmission were assessed. RESULTS: During the two-year study period, 1605 women underwent an index surgical admission. Among this population, a total of 177 readmissions (11.0%) in 135 unique patients occurred. In a surgical subpopulation with >1 night stay, a readmission rate of 20.9% was observed. The mean interval to readmission was 11.8days (SD 10.7) and mean length of readmission stay was 5.1days (SD 5.0). Factors associated with readmission included radical surgery for ovarian cancer (OR 2.87) or cervical cancer (OR 4.33), creation of an ostomy (OR 11.44), a Charlson score of ≥5 (OR 2.15), a language barrier (OR 3.36), a median household income in the lowest quartile (OR 6.49), and a positive discharge screen (OR 2.85). The mean cost per readmission was $25,416 (SD $26,736), with the highest costs associated with gastrointestinal complications at $32,432 (SD $32,148). The total readmission-related costs during the study period were $4,523,959. CONCLUSIONS: Readmissions to a high volume gynecologic oncology service were costly and related to radical surgery for ovarian and cervical cancer as well as to medical, socioeconomic and psychosocial patient variables. These data may inform interventional studies aimed at decreasing unplanned readmissions in gynecologic oncology surgical populations.

Association between days to complete inpatient discharge summaries with all-payer hospital readmissions in Maryland.

OBJECTIVE: Hospital discharge summaries can provide valuable information to future providers and may help to prevent hospital readmissions. We sought to examine whether the number of days to complete hospital discharge summaries is associated with 30-day readmission rate. PATIENTS AND METHODS: This was a retrospective cohort study conducted on 87,994 consecutive discharges between January 1, 2013 and December 31, 2014, in a large urban academic hospital. We used multivariable logistic regression models to examine the association between days to complete the discharge summary and hospital readmissions while controlling for age, gender, race, payer, hospital service (gynecology-obstetrics, medicine, neurosciences, oncology, pediatrics, and surgical sciences), discharge location, length of stay, expected readmission rate in Maryland based on diagnosis and illness severity, and the Agency for Healthcare Research and Quality Comorbidity Index. Days to complete the hospital discharge summary-the primary exposure variable-was assessed using the 20th percentile (>3 vs ≤3 days) and as a continuous variable (odds ratio expressed per 3-day increase). The main outcome was all-cause readmission to any acute care hospital in Maryland within 30 days. RESULTS: Among the 87,994 patients, there were 14,248 (16.2%) total readmissions. Discharge summary completion >3 days was significantly associated with readmission, with adjusted odds ratio (OR) (95% confidence interval [CI]) of 1.09 (1.04 to 1.13, P = 0.001). We also found that every additional 3 days to complete the discharge summary was associated with an increased adjusted odds of readmission by 1% (OR: 1.01, 95% CI: 1.00 to 1.01, P < 0.001). CONCLUSION: Longer days to complete discharge summaries were associated with higher rates of all-cause hospital readmissions. Timely discharge summary completion time may be a quality indicator to evaluate current practice and as a potential strategy to improve patient outcomes. Journal of Hospital Medicine 2016;11:393-400. 2016 Society of Hospital Medicine.

Decision support tool use in colorectal surgery: what is the role?

BACKGROUND: Decision support tools prioritizing transitional care can help decrease medical readmissions but little evidence exists within surgical specialties. MATERIALS AND METHODS: This study evaluated the use of early screen for discharge planning and discharge decision support system screening tools or selective multidisciplinary clinical evaluation for targeting post-acute care interventions among higher risk colorectal surgery patients based on 30-d readmission status. Patients with positive screening tool scores underwent standard discharge planning education and evaluation during index operation hospitalization and were referred for targeted post-acute interventions; patients with negative screening tool scores were further clinically evaluated for selective referral for post-acute interventions. RESULTS: We identified 300 colorectal surgery patients; 30.3% (n = 91) of patients had a positive screening score (early screen for discharge planning and/or discharge decision support system). Positive screening scores did not correlate with hospital readmission (35% of readmitted patients versus 29% of non-readmitted had a positive screen; P = 0.424). After negative screening scores, selective referral based on clinical assessment for postdischarge interventions helped to concentrate resources in patients who were later readmitted. Index hospitalization complications were significantly associated with positive screening tool scores whereas postdischarge complications were most predictive of readmission. CONCLUSIONS: Among colorectal surgery patients, selective clinical referrals appeared to be the best method for targeting post-acute interventions in patients at higher risk for readmission. Future research should focus on improving existing processes of care to reduce postoperative complications and constructing better tools to assess individual patients' needs for targeted interventions in the post-acute setting.