Preface Nurse Staffing and Inpatient Hospital Mortality Asked to discuss recent research on staffing and mortality Invited to expand discussion to other issues Jack Needleman, PhD FAAN Professor of Health Services, UCLA School of Public Health Associate Director, UCLA Patient Safety Institute Presented at 5 th Nursing Economic$ Summit June 7, 2012 2 for presentation Bending the cost curve/searching for value About to enter new period of cost containment in health care 1980s-1990s: Hunterization of cost containment Nursing as cost center, rather than service line New era perhaps more sophisticated Ambulatory: Accountable care s, medical homes Inpatient: Pay for performance, nonpayment for never events Nursing sensitive conditions as never events Res Implications Need for nursing to establish Service line Contribution to value for (business case) Need to change policy/payment to value what patients value 3 4 March 17, 2011 5 6
Why we did this study Address concerns raised about prior studies that questioned relationship of staffing and patient outcomes, including mortality: Cross-sectional studies comparing high and low staffed hospitals Open to alternative explanation of association with other factors correlated with staffing but not staffing Imprecise nurse staffing measurement Lack of adjustments for patient acuity Do not reflect intuitive sense of what short staffing is Alternative conceptualization and measure: Gap between actual and target staffing Compare actual staffing to target staffing Hospital established staffing targets: Unit specific Nursing care model Driven by census and patients need for nursing Other factors influencing nursing work load, such as patient turnover on unit Allows each to accommodate al nuances rather than fixed staffing targets applied to all s 7 8 We address these concerns by Constructing individual patient experience of low staffing based on day-to-day, shift-to-shift variations in staffing at the unit level Not institutional or unit annual or monthly average Same units, staff, technology, physicians We address these concerns by Constructing individual patient experience of low staffing based on day-to-day, shift-to-shift variations in staffing at the unit level Not institutional or unit annual or monthly average Same units, staff, technology, physicians Conducting study in a single institution that has: Lower-than-expected mortality High average nurse staffing levels Recognized for high quality by the Dartmouth Atlas, US News and World Report, and Magnet hospital designation Including extensive controls for potential sources of an increased risk of death Patient diagnosis and surgical status Patient demographics Unit admitted to 9 10 Data Sources Merged data from Patient classification/staffing system Patient classification data =>target/recommended staffing Staffing data =>actual staffing data Discharge abstracts, administrative data support system Patient demographics, diagnosis, discharge status Patient assigned unit by time during stay Anesthesia database to identify time not on assigned unit because in OR or other anesthesia related procedure Key measures of low staffing Below target shift Target staffing on a shift 8 hours or more below target 16% of all shifts Most likely on ICUs (but average hours/shift also higher) Evenly distributed across ICU shifts, more likely days and evenings on med-surg unit shifts Below-target shift High turnover shift Turnover for unit 1 standard deviation above day shift mean Turnover = (adm + disch + transfers)/census *2 7% of shifts Most on day, rare on night 11 12
Patient level measure of exposure to low staffing Cumulative count of below target and high turnover shifts Total count Count over first five days of Count over 6 shifts prior to discharge or death Use survival models for analysis with time varying covariates Allows us to incorporate information on changes in patient location (type of unit) and exposure to low staffing over Sample Large tertiary academic medical center 197,961 s over 2003 2006 43 inpatient units General care units (28) Step down (7) Intensive care units (8) 1,897,424 unit-shifts 13 14 Exposure to Below-Target & High Turnover Shifts Exposure to Number of Shifts with Actual Staffing Eight or More Hours Below Target Staffing, and to Number of Shifts with High turnover Exposure over first 30-days of % of Patients Exposure over first five days of % of Patients # shifts Below-target High-turnover Below-target High-turnover 0 319% 397% 343% 449% 1 197% 322% 214% 346% 2 137% 155% 149% 145% 3 92% 62% 99% 43% Increased Risk of Death With Exposure to Lower RN Staffing and Higher Patient Turnover Increased risk associated with each shift with RN staffing belowtarget or high turnover, 30 day cumulative exposure Shifts with RN staffing 8 or more hours below target Shifts with high patient turnover Increased risk associated with each shift with RN staffing belowtarget or high turnover, first 5 days cumulative exposure Shifts with RN staffing 8 or more hours below target Shifts with high patient turnover 2%/shift 4%/shift 3%/shift 7%/shift 15 16 Key findings Patient Mortality Even in a high quality hospital that generally meets its staffing targets and manages patient turnover, and extensive controls for the influence of other factors, we still could detect the effects of staffing and high patient turnover Effects are comparable to those observed in comparisons of high to low staffed hospitals Implications for Hospital Management No free passes for hospitals with high average staffing Need to strive to hit targets every shift Findings should also apply to hospitals less successful in routinely meeting nursing needs of patients Patients at higher average risk Operational implications Nursing service line, not just cost center Need systems for: Identifying target staffing Managing staffing against target Staffing for anticipated turnover Smoothing turnover 17 18
Challenges to applying within hospitals and to public reporting and research Standards for target staffing Basis for setting targets Opportunities for gaming or low-balling Measures Is 8 hours below target right for all units? How to incorporate patient turnover into analysis of adequate staffing Developing summary, cumulative measures of hospital experience Do we try to characterize and report each patient s experience Validation of measure with experience from other hospitals What do public reporting, regulatory and accreditation bodies do in response to these findings? 19 20 Factors leading to redesign of work Sicker patients Nursing shortage Changing technology with new opportunities, new demands EHR Smart beds and easier, more routine telemetry eicu Bending the cost curve will increase the scrutiny on nursing Under current payment systems and nursing models, increasing staffing to assure value to patients may not pay for hospitals Avoided Days and Adverse Outcomes Associated with Raising Nurse Staffing to 75 th Percentile Estimates from Needleman/Buerhaus, Health Affairs, 2006 RN Proportion Licensed Hours Do Both Avoided Days 1,507,493 2,598,339 4,106,315 Avoided Adverse Outcomes Cardiac arrest and shock, pneumonia, upper gastrointestinal bleeding, deep vein thrombosis, urinary tract infection 59,938 10,813 70,416 Avoided Deaths 4,997 1,801 6,754 21 22 Net Cost of Increasing Nurse Staffing Estimates from Needleman/Buerhaus, Health Affairs, 2006 RN Proportion Licensed Hours Both Cost of higher nursing $ 811 Million $ 75 Billion $ 85 Billion Avoided costs (full cost) $ 26 Billion $ 43 Billion $ 69 Billion Short term cost increase (save 40% of average) ($ 24 Billion) $ 58 Billion $ 57 Billion As % of hospital costs -01% 15% 14% 23 Factors leading to redesign of work Sicker patients Nursing shortage Changing technology with new opportunities, new demands EHR Smart beds and easier, more routine telemetry eicu Bending the cost curve will increase the scrutiny on nursing Under current payment systems and nursing models, increasing staffing to assure value to patients may not pay for hospitals Need to embrace quality-based payment and step up to claim ownership of task of improving care Redesign work for efficiency as well as improved safety & reliability 24
Challenges of redesigning work and sustaining change require nurses involvment Nonlinearity of work Ebright managing the stack Time? Completion and missed work? Accuracy/errors? Improving care requires integration of practices, not simply adopting best practices Need to sustain effective practices Not automatic Key roles in coordination invisible to patient and often other staff Gerri Lamb INQRI grant UCLA LOS and multi-disciplinary rounding Burden 25 26 Spaghetti diagram of nurse movement during 50 minutes of a shift Challenges to redesigning work and sustaining change require nurses involvment Nonlinearity of work Ebright managing the stack Time? Completion and missed work? Accuracy/errors? Improving care requires integration of practices, not simply adopting best practices Need to sustain effective practices Not automatic Key roles in coordination invisible to patient and often other staff Gerri Lamb INQRI grant UCLA LOS and multi-disciplinary rounding Burden Source, Institute for Healthcare Improvement, TCAB How-to Manual on Nurse Time in Direct Patient Care, 2008 27 28 We understand how to build a culture of improvement that engages the front line Leadership engagement and commitment Organizational commitment to safety and reliability Alignment of improvement goals and institutional goals Engagement, orientation and training of front line and clinical staff All above subject of pre-work in Michigan Keystone Weakening hierarchical relationships and empowering staff Respect the local wisdom of frontline providers Engaging front line staff requires addressing their concerns Adoption of methods for designing, testing and adapting innovations Plan-Do-Study-Act popular tool for rapid cycle testing Complements longer and more formal methods of analysis and designing innovation Commitment and capacity to collect and use data Not just culture but institutionalization of improvement work into work week and expectations 29