Multicenter Health System Research Articles

Integrated Care in Ontario: Unicorn or Black Swan?

The movement away from traditional models to organize, finance and deliver healthcare toward integrated models focusing on delivering value has been under way in many health systems and jurisdictions in the world with varying degrees of intensity and success for much of the past 20 years. I have had the opportunity to lead aspects of a multi-state health system committed to the concepts of accountable care during the first 10 years of the Patient Protection and Affordable Care Act (PPACA) in the US. For the past two years, I have assumed the role as CEO of a large academic health sciences centre in Ontario as the province embarks on a shift in policies to support integrated models of care delivery similar to those associated with the PPACA. I will describe my observations comparing two countries' move toward integrated delivery models and potential lessons for Canada.

The LACE + index as a predictor of 90-day urologic surgery outcomes.

This study assessed the ability of the LACE + [Length of stay, Acuity of admission, Charlson Comorbidity Index (CCI) score, and Emergency department visits in the past 6months] index to predict adverse outcomes after urologic surgery. LACE + scores were retrospectively calculated for all consecutive patients (n = 9824) who received urologic surgery at one multi-center health system over 2 years (2016-2018). Coarsened exact matching was employed to sort patient data before analysis; matching criteria included duration of surgery, BMI, and race among others. Outcomes including unplanned hospital readmission, emergency room visits, and reoperation were compared for patients with different LACE + quartiles. 722 patients were matched between Q1 and Q4; 1120 patients were matched between Q2 and Q4; 2550 patients were matched between Q3 and Q4. Higher LACE + score significantly predicted readmission within 90days (90D) of discharge for Q1 vs Q4 and Q2 vs Q4. Increased LACE + score also significantly predicted 90D emergency room visits for Q1 vs Q4, Q2 vs Q4, and Q3 vs Q4. LACE + score was also significantly predictive of 90D reoperation for Q1 vs Q4. LACE + score did not predict 90D reoperation for Q2 vs Q4 or Q3 vs Q4 or 90D readmission for Q3 vs. Q4. These results suggest that LACE + may be a suitable prediction model for important patient outcomes after urologic surgery.

Is Bariatric Surgery Safe and Effective in Patients with Inflammatory Bowel Disease?

The rate of obesity is rapidly increasing in patients with inflammatory bowel disease (IBD), but whether bariatric surgery in patients with IBD is safe and effective is not well understood. A retrospective review of patients with IBD undergoing bariatric surgery across a multi-state health system was performed. Thirty-day postoperative outcomes, weight loss, and long-term complications were recorded. Thirty-one patients (81% female) with IBD and a mean preoperative body mass index (BMI) of 42.4 kg/m2 underwent 32 bariatric operations (n = 14 Roux-en-Y gastric bypass, n = 14 sleeve gastrectomy, n = 4 gastric band). Short-term infectious complications included superficial surgical site infection (n = 2), infected intra-abdominal hematoma (n = 1), and a hepatic abscess (n = 1). Percent excess weight loss was 57.2% (n = 25) at 6 months, 62.9% (n = 22) at 12 months, and 57.4% (n = 11) at 24 months. No IBD flares requiring surgery were observed at a median follow-up of 2.7 years (interquartile range, 0.8-4.2 years). In carefully selected patients with IBD, bariatric surgery appears safe with respect to short-term infectious complications and results in sustained weight loss until at least 2 years postoperatively.

2169. An Algorithm-Based Approach Reduces Overuse of Meningitis/Encephalitis Multiplex PCR Panel

BackgroundSyndromic molecular panels enable rapid diagnosis and optimized management of infections with significant morbidity and mortality, but may be overused without clear guidelines. A recent report indicated there was little clinical suspicion of infection in up to 1/3 of cases for which a FILMARRAY® Meningitis/Encephalitis Panel (ME Panel, bioMérieux) was ordered. We recently implemented the ME Panel in our multicenter health system. We assessed ME Panel use for the 6-month period following test implementation.MethodsA testing algorithm was developed, vetted with our system-wide Infectious Diseases (ID) and Neuro-ID Services, and used as the basis for the education of the Emergency Medicine, Internal Medicine, Hospitalist, Pediatric, and Critical Care Medicine Services. Algorithm elements were embedded in the electronic medical record (EMR). Lab records and EMRs were reviewed for all patients tested by ME Panel or cerebrospinal (CSF) culture. Lab results, baseline demographic and underlying medical history, and results of singleplex viral PCR CSF tests and the multiplex NY State Encephalitis PCR Panel (NYS Panel, Wadsworth Laboratory, Albany, NY) were recorded. ME Panel results were compared with other findings.Results115 ME Panels were performed, with 5 (4%) positives [1 N.meningitidis, 1 H.influenzae, 1 cytomegalovirus (CMV), 1 Herpes simplex virus type 1 (HSV1), and 1 varicella zoster virus (VZV)]. Other findings were consistent with true infection for the N. meningitis, HSV and VZV; the CMV was likely reactivation. Significance of the H. influenzae was unclear. There were 830 CSF cultures, with 38 (4%) positive; 5 of the 38 were ME Panel targets. 29 NYS Panels were sent [1 positive each for Human Herpesvirus 6 (HHV6) and Epstein Barr Virus (EBV)]. Finally, 7 singleplex PCRs were positive [5 HSV, 1 CMV and 1 HHV6], including one negative by ME Panel.ConclusionWe did not find ME Panel overuse; rather, we found several cases for which the ME Panel could have given a more rapid diagnosis. We did identify areas for improvement in test ordering, such as minimizing duplicate testing (e.g., singleplex PCR) A multi-disciplinary approach engaging stakeholders in the lab, ID and Neuro-ID can assist appropriate test utilization and diagnostic stewardship.Disclosures All authors: No reported disclosures.

Implementation of a Preterm Labor Evaluation Algorithm in a Multi-State Health System [15L

INTRODUCTION: To determine the effect of implementing a standardized preterm labor (PTL) algorithm in a multiple hospital health system. METHODS: We performed a prospective evaluation of patients presenting with PTL at 34 hospitals within the Trinity Health System. Prior to implementation of a PTL algorithm, patients presenting with PTL symptoms were assessed in a manner determined by the clinician. After training, a PTL evaluation algorithm was implemented system-wide as a quality improvement project. The algorithm detailed the use of sterile speculum examination (SSE), fetal fibronectin (fFN) collection and transvaginal ultrasound cervical length (TVCL). Pre- and post-implementation data were compared using Chi-Square tests. Negative predictive value (NPV) was calculated for fFN and/or TVCL testing. RESULTS: 1,826 women (557 pre- and 1,269 post-implementation) were included in the quality improvement project. The rate of SSE (pre: 51%, post: 68%; P<.0001) and fFN performance (pre: 23%, post: 36%; P<.0001) increased after utilization of the algorithm, while the rate of TVCL remained unchanged (pre: 13%, post: 14%; P>.05). Maternal admission rates decreased with algorithm utilization (pre: 21%, post: 15%; P<.01). Women presenting with PTL symptoms who were not evaluated by fFN or TVCL delivered within 2 weeks at a rate of 24%, while those who received either fFN, TVCL, or both had a rate of 6% (P<.0001). NPV for fFN and/or TVCL diagnostic testing for delivery within 2 weeks is 96.8%. CONCLUSION: Use of the PTL algorithm improved the clinical evaluation of patients in our health system and decreased the maternal admission rate.

Neutropenia-Associated Outcomes in Patients With Breast Cancer Receiving Myelosuppressive Chemotherapy With Reduced Doses of Pegfilgrastim

Background: Pegfilgrastim can be utilized to prevent neutropenia-associated complications in patients receiving myelosuppressive chemotherapy for breast cancer. A common adverse event associated with pegfilgrastim is bone pain. Clinicians may opt to reduce the dose of pegfilgrastim when administering it to patients with previous severe or refractory bone pain. There is limited and conflicting evidence with regard to the safety and efficacy of this practice. The purpose of this study was to investigate the impact of administering reduced doses of pegfilgrastim on neutropenia-associated outcomes in patients with breast cancer receiving myelosuppressive chemotherapy. Methods: A retrospective chart review was conducted at a large, multistate health system with several different medical oncology practice sites. The primary outcome was the incidence of febrile neutropenia. Secondary outcomes included the incidence and severity of neutropenia, hospitalization for febrile, use of intravenous antimicrobials for febrile, delays in chemotherapy or dose reductions in chemotherapy secondary to neutropenia or febrile, rationale for dose reduction of pegfilgrastim, and improvement in bone pain. Results: A total of 80 patients received reduced dose pegfilgrastim. Most patients had their doses reduced secondary to bone pain (54%) or leukocytosis (14%). One (1.25%) patient experienced febrile neutropenia that did not require hospitalization or intravenous antimicrobials. Chemotherapy treatment delays and dose reductions secondary to neutropenia or febrile neutropenia occurred in 1 (1.25%) and 2 (2.5%) patients, respectively. Conclusion: Reduced doses of pegfilgrastim in patients receiving myelosuppressive chemotherapy for breast cancer resulted in a low incidence of neutropenia-associated events, including febrile neutropenia and grade 3/4 neutropenia.

Bridging the knowledge-practice gap in tuberculosis contact management in a high-burden setting: a mixed-methods protocol for a multicenter health system strengthening study

BackgroundPeople in close contact with tuberculosis should have screening and appropriate management, as an opportunity for active case detection and prevention. However, implementation of tuberculosis contact screening and management is limited in high-burden settings. Behaviour change is needed across three levels of the healthcare system—policymakers, healthcare providers, and patients. To bridge the wide policy-practice gap, this study draws on the Consolidated Framework for Implementation Research, the Behaviour Change Wheel, and the RE-AIM model (Reach, Effectiveness, Adoption, Implementation, Maintenance) to respectively understand barriers, implement change, and evaluate process and outcome.MethodsThis methods paper describes a mixed-methods intervention study in Eastern Indonesia. Quantitative data will be collected during baseline, intervention, and sustainability periods and analyzed using time series analysis. The primary outcome is the number of individuals completing tuberculosis preventive therapy by the end of the two-year intervention phase. Of an estimated 10,000 contacts during this period, we anticipate that a minimum of 416 will be found to have active TB or will complete preventive therapy. Qualitative data (semi-structured interviews, focus group discussions, and observations) will be collected from consenting healthcare providers, patients, and contacts. Activities to promote policy implementation include healthcare provider training, quarterly continuous quality improvement workshops, a social media discussion forum, and promotional materials. The Consolidated Framework for Implementation Research will be used to identify reasons for limited policy implementation at baseline. The Behaviour Change Wheel will be used to ensure that a suitable range of activities are implemented to facilitate change. The RE-AIM model will be used as the evaluation framework.DiscussionUse of theoretical frameworks in combination can ensure a comprehensive understanding of, and robust response to, health policy underimplementation. The selected frameworks are highly applicable to this pragmatic intervention study, in a setting where End TB Strategy targets will not be met without substantial behavior change within health systems. Continuous quality improvement cycles will provide a way to engage staff and stakeholders in understanding local data to motivate behavior change. If successful, up to 500 people could be prevented from developing complications of tuberculosis through early case-finding or receiving preventive therapy over a two-year period.Study registrationAustralian New Zealand Clinical Trials Register 375803.

Abstract TP373: Post IV-Alteplase Hemorrhage Rates in Early vs. Delayed Treatment Window in a Multi-State Health System

Introduction: Reported rates of symptomatic intracranial hemorrhage (sICH) following IV-alteplase (tPA) for acute ischemic stroke range from 2.4 to 6.4% in clinical trials, with even broader ranges in registries. The objective of this study was to examine rates of sICH following tPA, overall and in the <3 and 3-5 hour treatment windows, in a large health system ranging from critical access hospitals to comprehensive stroke centers, as there are sparse data comparing sICH rates by treatment window. Methods: Patients discharged from hospitals in a multi-state system between January 2009 and April 2018 with diagnosis of stroke and treated with tPA were selected for analysis. Patients with inpatient stroke, under age 18, receiving endovascular treatment or with incomplete data were excluded. Firth’s logistic regression was used to determine if patients receiving tPA in 3 to 5 hours from symptom onset were more likely to have sICH < 36 hours compared to those treated <3 hours from symptom onset, adjusting for age, admit NIHSS, and mode of arrival. Interactions were tested to determine if individual covariates had a moderating effect on sICH by treatment window. Results: There were 3,996 patient cases with complete data included in the analysis, with 17.8% (712) treated in the 3-5 hour window, 27 of whom were treated between 4.5 and 5 hours. Median NIHSS was 8.0 [interquartile range (IQR): 4.0, 15.0], median age was 72.0 [IQR: 60.0, 83.0], 49.0% (n=1,951) were female, and 78.6% (n=3141) arrived via EMS. The overall sICH rate was 3.3% (n=133). There was no difference in rate of sICH <36 hours by treatment window. Additional interactions also showed no difference. Conclusions: In a large health system, there was an overall low rate of symptomatic intracranial hemorrhage following alteplase administration and no increase in rate in the 3-5 hour treatment window.

Abstract TP399: Complications from IV Alteplase in Mild Stroke Patients in a Multi-state Health System

Introduction: The rate of complications from IV alteplase (tPA) for acute ischemic stroke is necessary for discussing risk and benefit when deciding whether to treat. Symptomatic intracerebral hemorrhage (sICH) rate is often quoted as 6.4% based on the NINDS trials. This risk often makes patients and physicians wary of tPA treatment for mild ischemic stroke (MIS). However, NINDS excluded these patients, so risk in MIS is not clear. The Objective of this study was to examine rates of tPA-related complications in MIS compared to moderate-to-severe stroke using real-world data. Methods: Data from a multi-state hospital system included tPA-treated patients discharged between January 2011 and May 2018. Patients who were <18 years old, received thrombectomy, transferred to another hospital, or had inpatient stroke were excluded. Outcomes included percentage of patients who experienced any complication, sICH, life-threatening systemic hemorrhage, other serious complication, or in-hospital mortality or discharge to hospice. Chi-squared and Fisher’s exact tests and generalized linear models were used to compare outcomes between patients with mild (NIHSS ≤ 5) and moderate-to-severe (NIHSS >5) strokes. Results: A total of 3,067 patients met inclusion criteria with 36% (1,111) having a mild stroke. When compared to MIS patients, moderate-to-severe stroke patients were more likely to have sICH (4.4% vs. 1.3%; p<.001), life-threatening systemic hemorrhage (0.8% vs. 0%; p<.001), and other serious complication (2.1% vs. 0.5%; p<.001). In multivariate analysis, patients with moderate-to-severe stroke were 3.69 (95% CI: 2.26, 6.01) times more likely to have any complication and 7.66 (95% CI: 4.82, 12.16) times more likely to die or be discharged to hospice after adjusting for age, sex, hospital, and history of atrial fibrillation, hypertension, diabetes, and coronary artery disease or prior myocardial infarction. Conclusion: Patients with mild strokes that received IV-alteplase had fewer treatment-related complications and were less likely to experience in-hospital mortality or discharge to hospice compared to those with moderate-to-severe strokes. This may inform decision-making in treatment of mild ischemic strokes.

Abstract TP357: Large Five State Health System Standardizes Stroke Patient Education While Significantly Improving Health Literacy

Background: Written stroke patient education materials are instrumental to improving the patient care experience. Since 2010, national certifying bodies recommend written materials at a maximum reading level of 5th grade to meet health literacy needs. In 2016 certified comprehensive and primary stroke centers from a multi-state health system discovered significant variation among stroke patient education booklets (SPEB) developed by separate facilities/states, and in 2018 implemented a standard SPEB by partnering with an external organization (EO). The health system worked with the EO to add information in the SPEB specific to carotid disease and stroke symptoms. Purpose: Compare the health literacy of pre-standardization SPEBs to the updated EO SPEB with and without carotid education. Methods: Readability assessments were used to determine reading grade levels for pre-standardization SPEBs consisting of 7 SPEB among 18 facilities in 5 states, the EO standardized SPEB with and without the additional carotid education, and the carotid education additions alone. Assessments included: 1) Flesch Reading Ease Formula, 2) Flesch-Kincaid Grade Level, 3) Gunning Fog Index Readability Formula, 4) Simple Measure of Gobbledygook (SMOG) Readability Formula, 5) Coleman-Liau Index, 6) Automated Readability Index, and 7) Linsear Write Formula. These assessments were then used to calculate a readability consensus score (RCS), reported as grade levels, using an online tool. Visual Imagery was not included in the assessments. Results: The range of RCS grade levels for pre-standardization materials was 7th to 12th grade. A grade level of 11 - 12 was found for 57.1% of sites. The EO SPEB without carotid education had an RCS grade level of 7 and with the additional information had a grade 8 level. The carotid education alone was at a college graduate level. Utilization of the EO SPEB decreased the health literacy level for 88.9% of sites. Conclusion: A standardized SPEB was implemented for a large five state health system. The health literacy level significantly improved at almost 90% of the participating facilities. The recommended 5 th grade health literacy level was not achieved based on the RCS. More research is needed on the impact visual imagery has on health literacy.

Implementation of breast cancer pathway for genetic counseling and testing in multi-state health system.

6521Background: Genetic counseling and/or testing (GC/T) are important aspects of breast cancer care. The National Comprehensive Cancer Network has guidelines for GC/T but reports demonstrate varia...

Abstract P260: Patients Experiencing Homelessness Have Longer Hospital Lengths of Stay After Admission for Ischemic Stroke or TIA

Introduction: There is little information available regarding stroke care and outcomes specific to patients experiencing homelessness. This study assesses the provision of stroke care and outcomes for these patients and identifies opportunities to improve their care. Methods: Data was obtained from a large, multi-state health system’s Get With The Guidelines registry for patients discharged between January 2009 and September 2016. Patients with an ischemic stroke or transient ischemic attack (TIA) were included. Patients with time of last known well (LKW) to hospital arrival of more than two weeks or hospital arrival to CT times of more than two days were excluded. Homeless patients were matched to non-homeless patients with a 1:5 ratio using case control matching on age, stroke type, gender and year. Patient characteristics between homeless and housed patients were compared before and after matching using t-tests, Mann Whitney tests, or chi-squared tests, as appropriate. Primary outcomes were discharge modified Rankin score (mRs), with disability categorized as none/slight (mRs≤2) or moderate or greater (mRS > 3) and length of stay (LOS). Logistic regression was used to analyze the relationship between homeless status and mRs, and Cox proportional hazards regression was used to assess the effect of homelessness on LOS. LKW to hospital arrival time and admit National Institutes of Health Stroke Scale (NIHSS) score were added as covariates to both models. Adjusted odds ratios (AOR) and hazards ratios (AHR), and their corresponding p-values were reported. Results: A total of 20,516 patients met inclusion criteria with 28 experiencing homelessness. Prior to matching, homeless patients were more likely to be male (78.6% vs 48.9%, p=.007), younger (58 vs 72 years, p<.001) and have longer median LKW to arrival times (921 vs 386 minutes, p=.066). After matching, the analytic sample size was 168 patients with 16.7% (n=28) experiencing homelessness and 83.3% (n=140) housed. Multivariable analyses indicated that homeless patients leave the hospital at a slower rate compared to the non-homeless (AHR=0.666, p=.056). They also had lower disability scores at discharge (AOR=0.104, p=.095); however, this result only approached significance. Conclusion: Patients experiencing homelessness had longer LKW to hospital arrival times indicating an opportunity for targeted intervention around the need for quick arrival after symptoms. In addition, homeless patients also had longer LOS which is likely related to challenges in safely discharging this population. Further research with a larger number of patients is needed in order to compare other aspects of acute stroke care, including treatment rates and time to treatment, and assess ways to address the challenges of shortening last known well to arrival times and safe discharge.

Medication utilization evaluation of dabigatran and rivaroxaban within a large, multi-center health system.

The objective of this medication utilization evaluation (MUE) was to determine the appropriateness of dabigatran and rivaroxaban while also reviewing outcomes for safety and effectiveness within a large, multi-center health system. A retrospective chart review was performed using the system's electronic medical record. A data inquiry was requested and generated for dabigatran usage from July 28, 2011 through July 28, 2012 and for rivaroxaban from March 1, 2012 to July 31, 2012 at eight health system hospitals. All patients receiving at least one dose were eligible for inclusion in the MUE. For dabigatran, 78 of 390 unique patient encounters were analyzed (20%). All 62 rivaroxaban encounters were included in the analysis. Dabigatran was used for appropriate indications in 94% of encounters and 82% for rivaroxaban. Based on indication and renal function, 87% of dabigatran patients and 92% of rivaroxaban patients received correct dosing. For patients transitioning to or from another anticoagulant, appropriate transitions occurred in 44% of dabigatran transitions and 48% of rivaroxaban transitions. At discharge, 83% of dabigatran and 86% of rivaroxaban therapy was continued. There were no reported strokes or systemic embolism with dabigatran, but one reported deep vein thrombosis occurred during hospitalization with rivaroxaban therapy. Documented bleeds in 5% of dabigatran and 3% of rivaroxaban patients. Patient education was documented for 37% of dabigatran and 26% of rivaroxaban patients receiving therapeutic anticoagulation. This MUE revealed the appropriate use of dabigatran and rivaroxaban therapy with few safety outcomes within a large, multi-center health system.

Abstract TP420: At-risk Stroke Patients in Need of In-hospital Lifestyle Interventions: Data From a Multi-state Health System

Introduction: Lifestyle factors such as weight, diet and smoking are associated with stroke incidence and recurrence. Targeted interventions on these risk factors can be successful in reducing stroke and TIA (Transient Ischemic Attack) rates. The objective of this study was to assess the number of patients presenting to a hospital with an ischemic stroke (ISC) or TIA who received lifestyle interventions in a large, multi-hospital health system. Methods: Data from 25 hospitals in a multi-state healthcare system stroke registry from 2002 to May 2015 were used. Patients with a diagnosis of ISC or TIA were included. Patients on comfort care, transferred to an outside facility from the emergency department, discharged to hospice or expired were excluded. Lifestyle intervention outcome measures included the percentage of patients who received information on weight and activity management, anti-hypertensive diet, cholesterol reducing diet and anti-smoking treatment. Outcomes were evaluated by risk groups including patients with Body Mass Index (BMI) ≥25, those prescribed anti-hypertensive or cholesterol lowering medications in-hospital and smokers. Results: A total of 39,200 patients met inclusion criteria. Analytic sample sizes ranged from 6, 576 to 39,200. Over sixty-four percent had BMI≥25 (n=16,527), 75.5% (n=23,364) were prescribed anti-hypertensive medication, 78.4% (n=29,609) were prescribed cholesterol lowering medication and 16.8% (n=6,576) were smokers. Among those with BMI ≥ 25, only 42.6% (n=6,560) received weight management intervention. Among those that were prescribed anti-hypertensive medication, 50.6% (n=8,463) received information on antihypertensive diet. Among those that were prescribed cholesterol lowering medication, 57.1% (n=10,648) received cholesterol lowering diet information. For those that smoked, 90.1% (n=5,923) received anti-smoking treatment. Conclusions: This large patient cohort demonstrates there are improvement opportunities for in-hospital stroke prevention efforts. While hospitals do a good job appropriately providing smoking treatment to stroke patients, lifestyle interventions such as weight management and diet should also be addressed for those at-risk.

Right Antibiotic for Right Bug: Impact of Criteria Based Indications on Reduction of Gram-Positive Agent Use at a Large Multistate Health System

Right Antibiotic for Right Bug: Impact of Criteria Based Indications on Reduction of Gram-Positive Agent Use at a Large Multistate Health System

No Harm, No Foul, and No Hospital Readmissions

No Harm, No Foul, and No Hospital Readmissions

Impact of Inpatient Harms on Hospital Finances and Patient Clinical Outcomes.

The aim of this study was to determine the impact of all-cause inpatient harms on hospital finances and patient clinical outcomes. A retrospective analysis of inpatient harm from 24 hospitals in a large multistate health system was conducted during 2009 to 2012 using the Institute of Healthcare Improvement Global Trigger Tool for Measuring Adverse Events. Inpatient harms were detected and categorized into harm (F-I), temporary harm (E), and no harm. Of the 21,007 inpatients in this study, 15,610 (74.3%) experienced no harm, 2818 (13.4%) experienced temporary harm, and 2579 (12.3%) experienced harm. A patient with harm was estimated to have higher total cost ($4617 [95% confidence interval (CI), $4364 to 4871]), higher variable cost ($1774 [95% CI, $1648 to $1900]), lower contribution margin (-$1112 [95% CI, -$1378 to -$847]), longer length of stay (2.6 d [95% CI, 2.5 to 2.8]), higher mortality probability (59%; odds ratio, 1.4 [95% CI, 1.0 to 2.0]), and higher 30-day readmission probability (74.4%; odds ratio, 2.9 [95% CI, 2.6 to 3.2]). A patient with temporary harm was estimated to have higher total cost ($2187 [95% CI, $2008 to $2366]), higher variable cost ($800 [95% CI, $709 to $892]), lower contribution margin (-$669 [95% CI, -$891 to -$446]), longer length of stay (1.3 d [95% CI, 1.2 to 1.4]), mortality probability not statistically different, and higher 30-day readmission probability (54.6%; odds ratio, 1.2 [95% CI, 1.1 to 1.4]). Total health system reduction of harm was associated with a decrease of $108 million in total cost, $48 million in variable cost, an increase of contribution margin by $18 million, and savings of 60,000 inpatient care days. This all-cause harm safety study indicates that inpatient harm has negative financial outcomes for hospitals and negative clinical outcomes for patients.