HIT Return on Investment Evaluating Progress in a

HIT Return on Investment: Evaluating Progress in a Sea of Change John Hsu, MD, MBA, MSCE AHRQ Conference 27 September 2007 1

HIT Background § Great potential for transforming clinical care, especially for patients with chronic diseases § Adoption of HIT across the U. S. is limited but growing § Actual benefits of HIT unclear: § § § Initial benefits of HIT depend on how routinely and systematically clinicians use the HIT tools and resulting information Little information on HIT effects in the ambulatory setting with commercially-available systems Actual benefits and costs of HIT are difficult to quantify § § Comprehensive identification Methodological challenges 2

Preliminary Results - IMPACT Study Impact of Information Technology on Clinical Care: An Evaluation of the Technology on Quality, Safety and Efficiency of Chronic Disease Care John Hsu, MD, MBA, MSCE (KP DOR) Ilana Graetz (KP DOR) Huihui Wang (KP DOR) Jie Huang, Ph. D (KP DOR) Mary Reed, Dr. Ph (KP DOR) Bruce Fireman, MA (KP DOR) Joseph Selby, MD, MPH (KP DOR) Yvonne Zhou, Ph. D (KP) Jim Bellows, Ph. D (KP CMI) Naomi Bardach, MD (UCSF) Julian Wimbush (UCB) Tom Rundall, Ph. D (UCB) Robert Miller, Ph. D (UCSF) Richard Brand, Ph. D (UCSF) Funding: AHRQ R 01 HS 015280 3

Overview • Design: – Longitudinal study with quasi-experimental changes in exposure to HIT, and using a pre-post analytic design with concurrent controls • Study Period: 2004 -2008 • Population: IDS Members with any of five chronic diseases in January 2004 (Asthma, CAD, DM, HF, Htn) • Data: - Automated databases Annual surveys 4

Basic HIT Tools CIPS First Available: e. Chart 1995 e. Rx/e. Refill e. Consult March 2004 Functions: Data-Review Ö Ö Ö Documentation Ö Order-Entry Ö Communication Paper-alternative: No Ö Yes Yes Integrated: Not integrated with other applications (i. e. , need log onto each application separately) Description: Viewing lab results Writing free-text visit notes Viewing medication list Requesting referrals or consultations Viewing medication list Using standard note templates Viewing medication allergies Sending messages to other providers Entering orders for new prescription or refills 5

KP Health. Connect Ambulatory Suite First Available: Staggered implementation (2005 -2008) Functions: Data-Review Ö Documentation Ö Order-Entry Ö Communication Ö Paper-alternative: No Integrated: Description: Fully Integrated • Viewing medication list, allergies, lab results • Using standard note templates & writing free-text visit notes • Order new prescription or refills with decision support • Ordering Disease-specific sets (drugs and labs) • Sending messages to other providers & requesting referrals or consultations • Sending and receiving messages from patients • Point-of-care access to decision-support tools– including: –Online references and resources for current treatment guidelines –Care Management Institute protocols, and standard tests/screens 6

Potential Benefits of HIT • Improved information availability (value of information) • Clinical benefits • Financial benefits: e. g. , greater efficiency, lower administrative costs, better coding Benefits predicated on clinician use of HIT tools 7

HIT Use

HIT Implementation and Use * Among office visits in department of Medicine or Family Practice 9

CPOE Implementation and Use * New prescriptions are defined as new prescriptions doctor wrote, can be refills for existing drugs or completely new drugs 10

HIT Use • Implementation ≠ use • Use of one type of HIT ≠ use of all HIT tools 11

Information Quality

Data Availability: Diagnoses Completed on Visit Date * Among office visits in department of Medicine or Family Practice 13

Clinical Benefits

Methodological Challenges for Assessing Clinical Benefits • Measures of use • Temporal trends - concurrent control groups • Patient- and physician-level differences • Reliable pre-implementation clinical data - differentiating documentation vs. care • Multi-level effects • Adequate power 15

Methods • Study Period: 04/2004 -12/2006 • Study Population – Active KPNC members who continuously enrolled during the study period – 18 years and older as of 04/01/2004 – In diabetes registry as of 1 st quarter of 2004 – Members in 5 medical centers where KPHC implemented before 07/2006 during the study period – In teams which existed all the time during the study period – With at least one LDL measurement in pre-HIT period and one in post-HIT period • Predictor Measures: Presence of HIT (Health. Connect) • Model: Mixed model with random effects at PCP and Patient level, adjusted for patient age, gender, race/ethnicity, neighborhood SES, time of measurement and Medical centers 16

Definitions of Presence of HIT Definition 1: Medical center level KPHC rollout schedule – HIT=0: before KPHC was implemented at the first team in the medical center – HIT=1: within six months after KPHC was implemented at the first team in the medical center – HIT=2: six months after KPHC was implemented at the first team in the medical center Definition 2: Primary care team level actual use – HIT = 0: low use (<80% at team level) of e. Chart or KPHC – HIT = 1: starting from the first month when e. Chart used >=80% – HIT = 2: starting from the first month when KPHC used >=80% 17

Mean LDL in Each Month in KPNC 18

Association between HIT and LDL Estimate 1. Implementation at Before KPHC Medical Center (rollout schedule) First 6 months of KPHC 6+ months of KPHC 1. 00 ref. group -0. 50 -1. 15 0. 15 -0. 64 -1. 58 0. 30 Low HIT use (<80% of visits) 1. 00 ref. group EChart used in >80% of visits -0. 89 -1. 55 -0. 23 KPHC used in >80% of visits -1. 72 -2. 68 -0. 76 2. Actual use by Primary Care Team (% of total visits) 95% CI 19

Costs

Investment • Investment costs – Equipment – Personnel/productivity – Training • Maintenance costs – IT support staff – Future upgrades – Continued training 21

Other Relevant Features

Dynamic Environment • Changes in HIT – Decision support – Information use • Changes in Care Delivery – Clinical coordination – Delivery system • Changes in Medical Therapy – Information on effectiveness – Dissemination of new knowledge • Changes in the Market – Payment features, e. g. , risk adjustment, reporting, performance incentives – Payment mix 23

Conclusions • Benefits – Some potential clinical benefits related to better information at the point-of-care – Unclear benefits associated with improvements in clinical information at the system level – Transaction benefits perhaps easiest to quantify – Financial benefits depend market and reimbursement mix • Costs – Investment costs beyond equipment costs can be difficult to quantify – Maintenance costs also important • Dynamic/changing systems and markets. . 24

Summary: Need for Better Empirical Studies 25

HIT as Basic Infrastructure 26