Leveraging Electronic Health Records to I Improve Hospital H i l Performance: P f The Role of Management Academy Health ARM June 8, 2014 Julia Adler-Milstein, Kirstin Scott, Ashish Jha The EHR EHR-Performance Performance Paradox On the one hand… On the other hand… Studies from individual institutions reveal substantial quality and efficiency gains from EHRs Growing number of national studies that fail to find a consistent relationship between EHR adoption and improved performance Key question K ti iis: under de what hat co conditions ditio s do EHRs lead to improved performance? The EHR EHR-Performance Performance Paradox Evidence is beginning to emerge on what else is required Dranove, Forman, Goldfarb, Greenstein (2012) Examine the relationship between EHR adoption and operating costs Key findings Time effect: short-term increase Conditions matter: long-term g decrease for some;; increase for others Availability of technology skills in the local labor market Internal IT expertise appears to have little impact on the relationship between EHR adoption and costs What other types of organizational competencies help ensure that EHRs performance improvement? The EHR EHR-Performance Performance Paradox Degree to which an organization is well-managed is likely to be an important facilitator of IT-enabled performance improvement IT is a tool that requires concerted and thoughtful application Without this, IT can make performance worse by introducing complexity l i that h organization i i iis not equipped i d to manage Research R h Question: Q ti D Does management quality li modify dif the relationship between EHR adoption and cost and quality outcomes? Data Hospital HIT adoption: AHA IT supplement (2009) Hospital characteristics: AHA annual survey (2009) Hospital p management: g From Sadun,, Bloom,,Van Reenen ((2010)) Hospital outcomes: MedPAR (2010) Data – AHA AHA IT Supplement is administered to all U.S. hospitals annually Asks about adoption of various IT functionalities C ll t d M Collected March h – September S t b 2009 2009, with ith 69% response rate t Measure: Hospital had the functionalities that comprise at least a “basic” EHR EHR adoption Results viewing CPOE for f medications di i Clinical documentation AHA Annual Survey Additional hospital characteristics: size, teaching status, ownership, system-affiliated, urban location, & % Medicare Data – Management Random sample of 325 AHA hospitals (2009) Double-blind phone interviews Typically department or unit managers in cardiology or orthopedics Validated tool with 20 questions scored on 1 (poor) to 5 (excellent) scale 4 dimensions of hospital management: Operations: do companies use techniques like lean manufacturing, and do they document process improvements and the rationale behind introductions of improvements? Monitoring: M it i h how wellll d do companies i monitor it what h t goes on iinside id their th i fifirms and d use thi this for continuous improvement? Targets: do companies set the right targets, track the right outcomes, and take appropriate i action i if the h two are iinconsistent? i ? People: are companies promoting and rewarding employees based on performance, and trying to hire and keep their best employees? Data – Hospital Performance (AMI) For all fee-for-service Medicare patients with a primary diagnosis off AMI in 2010, 20 0 we calculated the following f riskadjusted hospital-level measures: Length-of-stay (LOS) 30-day mortality Payment per discharge Analysis Analytic sample: 191 U.S. acute-care hospitals in both management sample and 2009 that responded to AHA IT supplement in 2009 C Cross-sectional i l analyses l (OLS) (OLS), with i h non-response weights i h Base Model: Average management score, EHR, and hospital controls Interacted Model: with EHR*management * score interaction Robustness Tests: I Interview-level i l l controls l Individual EHR components Individual management components Graph predicted margins at 0.50 increments of management score Summary Statistics: EHR Adopters vs vs. Non Non-Adopters Adopters At least basic No basic EHR EHR (N=23) Mean (N=168) Std Std. Mean Dev. Std Std. P P- Dev. Value Risk-Adjusted Length of Stay (days) 5.13 2.30 4.86 1.96 0.599 AMI 0.19 0.21 0.19 0.17 0.971 14,440 8,394 12,298 7,268 0.264 3.18 0.65 2.99 0.53 0.191 OUTCOMES FOCAL PREDICTOR 30-day Mortality Payment per Discharge ($) Management Composite Score Summarized OLS Regression Results OUTCOME: MODEL: EHR MGMT EHR X MGMT LENGTH OF STAY 30-DAY MORTALITY PAYMENT Base Interaction Base Interaction Base Interaction (1) (2) (3) (4) (5) (6) 0.011 4.62* 0.03 0.19 184.59 24,417.89** [0.50] [2.55] [0.05] [0.20] [1,968.95] [10,329.61] -0.42 -0.14 -0.05 -0.04 -508.82 976.70 [0.30] [0.30] [0.03] [0.04] [1,207.09] [1,021.58] -1.48* -0.05 -7,786.74** [0.77] [0.06] [3,140.69] Coefficients C ffi i t from f OLS regressions i th thatt iinclude l d hhospital it l characteristics. h t i ti Brackets contain robust standard errors; *** p<0.01, ** p<0.05, * p<0.10 Predictive Margins: Length of Stay EHR No EHR 9 Poorly Managed: EHR – 6.9 No EHR – 5.0 8 7 Day ys 6 5 4 3 Well Managed: EHR – 3.7 No EHR – 4.7 2 1 0 1 1.5 2 2.5 3 3.5 Management Score 4 4.5 5 Predictive Margins: Mortality EHR No EHR 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 1 1.5 2 2.5 3 3.5 Management Score 4 4.5 5 Predictive Margins: Payment Days s EHR No EHR 9 8 7 6 5 4 3 2 1 0 1 1.5 2 2.5 3 3.5 Management Score 4 4.5 5 Limitations Small sample p Not all management data from cardiology units Associations not causal relationships Associations, Pre HITECH & meaningful use Summary – Key Findings Consistent with other studies,, no direct relationship p between EHR adoption and hospital performance However, the quality of the management appears to modify the relationship between EHR adoption and hospital performance For efficiency related measures: LOS and payment Some evidence for mortality (likely underpowered) Suggests that improving hospital management may be a critical f factor to ensure that h EHRs EHR iimprove hhospital i l productivity d i i Summary – Potential Mechanisms Motivation for adopting EHRs… Well-managed g Poorly-managed g Performance improvement External pressures (e.g., HITECH) Relationships with frontline Engaged partners staff… Top-down, disconnected Use of advanced functionalities… Limited – “basic” use only Clinical decision support in pparticular Regular, ongoing Use of EHR data for pperformance measurement measurement and monitoring… Sporadic or none Thank you!