Automated Telephone Self-management Support:

Automated Telephone Self-management Support: A Public Hospital Innovation with Great Potential Dean Schillinger, MD Professor of Medicine Chief, CA Diabetes Prevention and Control CA Dept of Public Health UCSF Center for Vulnerable Populations San Francisco General Hospital Margaret Handley, MPH PhD Assistant Professor of Epidemiology

Rationale: What is Health Literacy? The degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make [informed] health decisions. -Institute of Medicine, 2004

1 st National Assessment of Health Literacy n=19,714 Below Basic: Circle date on doctor s appointment slip Basic: Give 2 reasons a person with no symptoms should get tested for cancer based on a clearly written pamphlet Intermediate: Determine what time to take Rx medicine based on label Proficient: Calculate employee share of health insurance costs using table National Center for Educational Statistics, U.S. Department of Education, 2003

1 st Health Literacy Assessment 12% n=19,714 U.S. Adults Proficient 53% Intermediate Below Basic 14% Basic Hispanic 22% Average Medicare National Assessment of Adult Literacy (NAAL): National Center for Educational Statistics, U.S. Department of Education, 2003.

Patients with Diabetes and Low Literacy Less Likely to Know Correct Management Need to Know: symptoms of low blood sugar (hypoglycemia) Low Moderate e High Need to Do: correct action for hypoglycemic symptoms Low Moderate High 0 20 40 60 80 100 Percent *Williams et al., Archive of Internal Medicine, 1998 Williams 1998

Literacy is Associated with Glycemic Control, N=408 50 Inadequate Marginal % of patien nts 40 30 20 Adjusted OR=0.57, p=0.05 Adjusted OR=2.03, p=0.02 Adequate e 10 0 1st Quartile (Tight Control: HbA1c 7.2%) 4th Quartile (Poor Control: HbA1c>9.5%) Schillinger JAMA 2002

Adjusted odds of self-reported diabetes complications, for patients with inadequate vs. adequate literacy (N=408) Complication n ** AOR 95% CI Retinopathy 111 2.33 (1.19-4.57) Nephropathy 62 1.71 (0.75-3.90) Lower Extremity Amputation 27 2.48 (0.74-8.34) Cerebrovascular Disease 46 2.71 (1.06-6.97) Ischemic Heart Disease 93 1.73 (0.83-3.60) Schillinger JAMA 2002

Schillinger 2004 Diabetes Patients with Limited Literacy Experience Poorer Quality Communication, N=408 50 Inadequate FHL Adequate FHL 40 OR=3.2;p<0.01 01 OR=1.9;p=0.04 30 OR=3.3;p=0.02 OR=2.4;p=0.02 20 32% 26% 33% 21% 10 20% 13% 13% 13% 0 Doctor Use Words Give You Test Results Confused About Doctor Understand Not Understood w/o Explanation Medical Care Problems Doing Rx (Often/Always) (Often/Always) (Often/Always) (Never/Rarely/ Sometimes)

The Impact of Language Barriers on Poor Glycemic Control Among Insured Latino Diabetics: Data from DISTANCE Study Fernandez A, Schillinger D, Warton M, Parker M, Adler N, Schenker Y, Moffet H, Salgado V, Ahmed A, and Karter A.

Glycemic Control of Latino Diabetics by English Language Ability and by Physician-patient Language Concordance English Speakers (n=2683) All LEP (n=510) P Value LEP - LC (n=137) LEP- LD (n=115) P Value A1c, mean (SD) 7.65 (1.71) 7.81 (1.85) 0.06 7.58 (1.62) 7.99 (1.92) 0.07 Proportion of group with A1c 9% (%) 18.0 21.4 0.08 16.1 27.8 0.03 Abbreviations: PCP: Primary Care Physician; LEP: Limited English Proficient; LEP-LC: LEP with language concordant PCP; LEP-LD: LEP with language discordant PCP

80% 70% 60% 50% 40% 30% 20% 10% 0% Self-Management Support and 69% telephone 55% group visit Communication 42% internet Interest in Self-Management Support Research questions 1. Are vulnerable diabetes patients interested in selfmanagement support? 2. Do they perceive a benefit to improved communication? 40% perceived a benefit from better provider communication, higher for racial/ethnic minority and limited it health literacy (70%) Conclusion: Vulnerable diabetes patients t desire self-management support and perceive a benefit to improved communication. Sarkar et al., Patient Education and Counseling 2008

IDEALL Project: Improving Diabetes Efforts Across Language and Literacy Community Health Network of SF/DPH AHRQ CMWF, TCE, CHCF Schillinger Diab Care 2009

Automated Telephone Diabetes Self- Management (ATSM) Nurse Diabetes Care manager ATSM: Weekly Monitoring and Health Education Primary Care Physician Patient Interactive health technology, touch tone response Weekly surveillance & health education (39 weeks=9 mos) In patients preferred language (English, Spanish or Cantonese) Generates weekly reports of out of range responses Live phone follow-up through a bilingual nurse ->behavioral action plans

Components of Intervention Proactive, outgoing automated t calls; interactive ti health technology, touch tone response. 6-8 minutes/call (weekly surveillance & health education [39 weeks=9 mos] ) In patients preferred language (English, Spanish or Cantonese) Rotating/recurring topics of questions and health education narratives re healthy eating, exercise, medication adherence, coping/mood, self-monitoring, etc. Generates weekly reports of out of range responses based on a priori thresholds we set For those who trigger : Live phone follow-up through a bilingual nurse ->behavioral action plans/problemsolving

Components of Intervention About 50-60% of ATSM calls get answered and completed About 40-50% of these have triggered a call-back. Care manager able to reach the vast majority for f-u and develop patient-generated action plans. Most patients achieve partial or complete success on AP To date, care manager not empowered to up-titrate medications; does communicate with PCP re safety issues or significant ifi problems. Care manager has database for ease of data entry; currently not integrated into medical EHR

Key Findings of IDEALL Program Estimating Public Health Reach of Programs Composite reach product ATSM Group Visits (GMV) Overall 22.1 4.8 English 20.00 64 6.4 Chinese 22.0 2.7 Spanish 24.3 4.0 Adequate Literacy 15.6 7.6 Limited Literacy 28.0 3.6 Schillinger, et al.health Ed and Behavior 2007

Results: Structure and Process Measures 70 60 * * 58.9 60.2 50 40 48.2 41 30 36.8 39.3 20 UC ATSM GMV PACIC 75 * 70 72.9 65 68.9 65.4 60 62.9 63.4 55 59.2 50 UC ATSM GMV Communication pre post 80 75 70 65 60 5 4 3 *P<.05. * * 77.2 77.2 73.5 73.3 71.7 71.7 UC ATSM GMV Self-Efficacy * 4.4 3.9 3.8 3.9 3.7 UC ATSM GMV * 4.1 Self-Management Behavior Schillinger, in press Diabetes Care

Results: Functional Outcomes 5 4 3 2 1 0 Rate ratio 0.5 vs UC, 0.35 vs GMV 3.9 3.8 3.6 3.1 * 1.4 UC ATSM GMV Bed Days 3.6 pre post 20 15 10 5 0 17 13 14 OR 0.37 vs UC 18 UC ATSM GMV Diabetes Interference * 6 17 70 65 60 55 50 58.8 64.2 57.2 67 61.7 63 65 60 55 50 45 60.2 56.7 50 51.3 50.9 57.1 UC ATSM GMV UC ATSM GMV SF12 - Mental Health SF12 - Physical Health *P<.05

Results: Physiologic Outcomes 145 80 140 135 141.5 139.6 136.9 137.1 142.4 138.9 75 78.1 78.5 75 78.1 75.4 75.5 130 70 UC ATSM GMV SBP pre post UC ATSM GMV DBP 10 9 8 7 9.8 9.3 9.4 9 9 8.7 UC ATSM GMV 33 32 31 30 29 28 32.4 32.1 31.2 31.4 30.3 30.7 UC ATSM GMV HbA1c BMI

ATSM as Surveillance Tool? ATSM Data Automated Completed Calls Patient-Nurse Encounters CONSENSUS AE PotAE No event Medical Record Classification - Preventability - Primary Provider Awareness

Automated telephony provides safety surveillance function 120 111 participants, p 54% inadequate 100 health literacy 80 264 events 60 among 93 40 participants (86%) 20 111 AE s and 153 PotAE s 0 Sarkar, Schillinger et al. 2008 JGIM Number of Events Preventability Incident AE Prevalent AE PotAE Incident Prevalent Preventable Ameliorable PotAE Unable to determine Non-preventable

Clinician Survey Findings Responses from 87 of 113 (77%) physicians who cared for 245 of the 330 (74%) patients (mean, 2.8 per physician). Compared to UC, patients exposed to ATSM were perceived as more likely to be activated to create and achieve goals for chronic care (standardized effect size, ATSM vs. UC, +0.41, p=0.05). 05) Over half of physicians reported that ATSM helped overcome 4 of 5 common barriers to diabetes care Physicians rated quality of care as higher among patients exposed to ATSM compared to usual care (OR 3.6, p=0.003), and compared to GMV (OR 2.2, p=0.06) The majority felt ATSM should be expanded to more patients with diabetes (88%) a technology-facilitated SMS model was particularly effective for their patients and practice settings, suggesting that such programs should be disseminated and implemented more widely. Bhandari, Schillinger SGIM 2008

Health System Findings: Cost-Effectiveness; Health Plans Based on functional improvements, we estimated that the cost per QALY for ATSM was: >$65,000 for both set-up and ongoing costs >$ 32,000 for ongoing costs only Cost effectiveness could be further improved with (a) scaling up or (b) metabolic outcomes improved A large majority of CA Medicaid health plans reported an interest in employing ATSM-like technology Handley, Schillinger, in press Ann Fam Med 2008 Goldman, Schillinger et al. Am J Man Care 2007

Key Findings of IDEALL Program Reach significant, ifi especially for lower literacy, non- English speaking, Medi-Cal, uninsured. Interactive health technology improves patient centered care, health behaviors, functional status and promotes safety, due to proactive nature hierarchical logic communication tailoring For physiologic i effects to be achieved, need medication intensification Health plans and clinicians favorably inclined A challenge for individual clinics to implement

Current Project Partner with a local Medicaid health plan: San Francisco Health Plan SFHP care managers will make ATSM response calls Test effectiveness when implemented in realworld Compare ATSM-ONLY with ATSM-PLUS (medication activation) ATSM-PLUS involves merging pharmacy claims data with ithatsm data to enable care manager counseling

Design and Outcomes Wait List Design, with randomization among exposed participants. i t Total N=500 Outcomes (wait-list vs. ATSM vs. ATSM-Plus): -communication -behavior -functional status -metabolic indicators -patient safety (prevalence and root causes)

Suboptimal Refill Non- Self-Reported Goals on Adherence Med Non- Diabetes on Pharmacy Adherence on Registry Claims ATSM SFHP Care Manager Call to Patient: Check understanding and educate regarding diabetes goals Elicit barriers to adherence Inform about current data & goals Assess understanding of discussions with PCP Assess willingness to increase or add new medication to meet goals Develops action plan using motivational interviewing principles

SFHP Pre- Enrollment Post Card English

Spanish

Cantonese

SFHP Wallet-Size Card English, Spanish and Cantonese

Care manager field

Potential Safety Event

Safety event assessment

Engagement g with Smart Steps Sample: 186 SFHP patients enrolled in Smart Steps in 2009-2010 actively receiving calls in March, 2010, who had completed at least 3 weeks of calls. Results: Overall, 132 of 186 (71%) engaged with program This represents: 89% engaged among Cantonese (94/106) 86% engaged among Spanish (12/14) 40% engaged among English speakers (26/66)

Improvements/Threats Improvements to future dissemination: Care manager health coach Harnessing pharmacy claims data Marketing and outreach Trusted health plan Potentially sustainable Development of detailed training manual/qa processes Threats to implementation: Delays in implementation Staff turnover @health plan Maintaining fidelity to intervention processes Care mgr processes; claims data/registry data incomplete Coordinating treatment preferences/medication activation with PCP