Studying TBI Using a Natural Experiment Approach Kaiser Foundation Rehabilitation Center, Vallejo, CA Kabat-Knott Center for Rehabilitation Research

Transcription

1 Studying TBI Using a Natural Experiment Approach Kaiser Foundation Rehabilitation Center, Vallejo, CA Kabat-Knott Center for Rehabilitation Research M. Elizabeth Sandel, MD Chief, Physical Medicine and Rehabilitation Napa Solano Service Area, KP Northern California

2 TBI Studies: A Natural Experiment Approach Understand the potential of an electronic medical record to compare TBI populations within an integrated health system Describe the usefulness of variation in care within a health system to study TBI outcomes Define the most available and useful variables for TBI research that might be analyzed using this approach Identify various types of care, care settings, and care delivery for TBI that might be compared using this approach 2

3 KP Health System: Integration Northern California: 3.2 million members US: 8.6 million members Prepayment Non-Profit Health Plan & Hospitals (36) + other Medical Centers Permanente Physicians (Salaried) Federation of Permanente Medical Groups Pharmacies and Labs at the Medical Centers IT and data systems: EMR throughout the national system inpatient and outpatient Quality Operations and Support Divisions of Research in most Regions 3

4 KFRC at VAL MC Contract Facility SCVM Contract Facility RCMC

5 Kaiser Foundation Rehabilitation Center and Hospital Vallejo, California: Opened March 2010 Page 5

6 Neurotrauma Northern California: 3.2 million members SCI: 130 = approximate annual incidence TBI: 6000 = approximate annual incidence TBI Registry (Mild TBI): suggests the incidence may be higher 80,000+ MR# s over 6 years 6

7 Evidence Translation: 3 Tiers (AHRQ) T1: Clinical Efficacy: Translating basic science into clinical efficacy T2: Clinical Effectiveness: Using patient-oriented outcomes and health services research to develop knowledge about clinical effectiveness T3: Implementation Research: Using implementation research for continuous measurement and refinement of treatment implementation 7

8 TBI Research We are just beginning our work in TBI with a project on TBI and Alzheimer s Disease. TBI research will be based on our work in stroke Retrospective studies Prospective studies Variation in care explorations Implementation research 8

9 Paradigm: Post-Acute Care Stroke Studies Retrospective Studies 1. Disparities in Stroke Rehabilitation (CDC) 2. Stroke Mortality in Post-Acute Care (KP Community Benefits) 3. Stroke Onset-Days (KP Community Benefits) Prospective Studies 1. Outcomes Monitoring Study (NIH Clinical Center) 2. OMS Sub-study (NIH-CC) 3. KP-Functional Outcomes System (NINDS) 9

10 Research: Post-Acute Care after Stroke Retrospective Studies 10

11 Research Questions: Disparities Study What are the referral and enrollment rates for rehabilitation (post-acute care) following discharge from the hospital after a stroke? Are there disparities in care for patients in various settings based on variables such as: Race/ethnicity Gender Age Socioeconomic status: education and income level Hospital referral patterns Rural/urban setting Type of stroke 11

12 Research Methods: Disparities Study Retrospective cohort study within the Northern California Kaiser Permanente Health System Methods: Tracked rehabilitation services for 365 days following acute hospitalization for a first stroke Participants: 11,119 stroke patients hospitalized for acute stroke from Outcome measures (service delivery): Inpatient rehabilitation hospital, skilled nursing facility, home health services, outpatient rehabilitation services Variables: Age, gender, race/ethnicity, socioeconomic status (zip code), rural/urban residence, medical center referrals, type of stroke 12

13 Population Characteristics: Disparities Study Mean age: 69.7, 51% female Race/ethnicity: 70.2% white, 10.8% black, 8.8% Asian, 7.4% Hispanic Median household income: $56,750 Median percentage with at least a high school education: 87%; 28% had 4 years or more of college education Type of stroke: 85% ischemic, 15% hemorrhagic LOS: Median = 3; mean = 5.2 White population was older at the time of stroke Higher percentages of white and Hispanic populations were living in rural areas Asians were more likely to have a hemorrhagic stroke Median household incomes were lower for black and Hispanic populations 13

14 Results: Disparities Study Post-acute care population: increased over time as the membership population increased Percentage discharged directly to IRH and home increased SNF and HH care decreased and outpatient care increased Age: Younger patients more likely go to IRH; older patients to SNF Gender: Female patients more likely to receive IRH and SNF care Race/ethnicity: Asian and black patients more likely to receive IRH care than whites and Hispanics Geography: Central Valley and Capital Service areas less likely to receive IRH and SNF care; rural patients less likely to receive less IRH, HH, and more likely to have only OP care or no care at all Socioeconomic status: Patients with higher income and more education more likely to go to IRH; patients with lower income more likely to go to SNF 14

15 Research Questions: Stroke Mortality in PAC Do mortality rates vary with populations receiving different types of post-acute services? Does the prescription of anti-platelets, anticoagulants, cardiac and antihypertensive drugs, lipid-lowering agents, diabetic medications, and antidepressants link to a reduction in mortality? Are there disparities or variations in stroke mortality based on age, gender, race/ethnicity, socioeconomic status, rural/urban residence, service area, or other factors? 15

16 Findings: Stroke Mortality in PAC Post acute care services within 14 days of acute care hospital discharge were grouped into IRH, SNF, home health, outpatient rehabilitation treatment (hierarchy of care based on service intensity) SNF is associated with a higher mortality rate than other PAC services; IRH linked to better survival rate than HH in 14 day PAC grouping Medication prescription: associated with better survival Older age, male gender, African-American race, previous stroke, a longer acute care hospital stay - linked to a higher follow-up mortality rate. 16

17 Research Questions: Stroke Onset Days Does the duration of the onset-days interval from acute care discharge to admission to IRH care affect the functional outcome of stroke patients receiving inpatient rehabilitation treatment? What is the optimal onset-days interval that results in the best functional outcomes of stroke patients? Are there other influential factors for functional outcome (FIM), e.g., age, gender, race/ethnicity, socioeconomic status, and rural/urban residential area, clinical and functional measures at IRH admission, medication use and IRH length of stay. 17

18 Findings: Stroke Onset Days Functional recovery of patients with moderately to severely impaired stroke appears to depend in part on the timeframe from acute stroke to IRH (within 21 days). Also of importance to functional outcome is the prescription of medications to manage such co-morbid conditions as depression, cardiovascular disorders, hypertension, and diabetes. Age, race/ethnicity, type and side of stroke, history of a previous stroke, initial functional measures at IRH admission, and IRH length of stay are linked to the functional improvement of stroke patients. 18

19 Research: Post-Acute Care after Stroke Prospective Studies

20 Multiple Sites of Post-Acute Care Sites of Care 2 5 % Receiving Each Pattern of PAC IR F S N F H H O P IR F S N F H H IR F S N F O P IR F S N F IR F H H O P IR F O P IR F S N F H H O P S N F H H S N F O P S N F H H O P H H O P N O N E 20

21 What Key Data Elements are Missing from the EMR Data Bases for Stroke? Severity of illness, injury, or condition (NIHSS*) Functional status (AM-PAC*) Activity Participation in social roles Social and Family Data (added to the AM-PAC*) Socioeconomic data Education Living situation.so, *these data elements were added into our prospective study.

22 Functional Assessment Activity Measure for Post-Acute Care (AM-PAC) Developed at Boston University by Jette, Haley, et al Used in our prospective studies in stroke Potential data elements to be used in retrospective studies if it is built into the clinical and research data bases and data is entered by patients or proxies Can be supplemented with additional questions.

23 AM-PAC and CAT Logic AM-PAC was originally developed as a standard assessment instrument using a structured interview to locate a patient s function in a domain AM-PAC developers used Computer Assisted Testing (CAT) logic to establish a computer-presented assessment that presents questions based on the patient s response CAT logic reduces test time because it only presents questions that are close to the patient s current function CAT methodology eliminates ceiling and floor effects and is therefore ideal for studying patients across a care continuum 23

24 AM-PAC CAT Overview The application delivers questions about a patient s function in five domains: 1. Mobility 2. Activities of daily living 3. Applied cognition 4. Social participation 5. Role participation An AM-PAC CAT session takes about minutes 24

25 A Sample AM-PAC CAT Item How much DIFFICULTY do you currently have bending over to pick up something from the floor without holding onto anything? Unable A Lot A Little None Depending on the answer, the CAT logic selects another question. For example, if the answer is None, the CAT logic selects a question for a more difficult task. 25

26 AM-PAC or ecat Functional Outcome Report Patient Facility Episode Visits Admission Bending over: A Lot Difficulty Walking outdoors: A Little Difficulty Stand from chair: A Lot of Difficulty Walking indoors: A Little Difficulty Discharge Bending over: No Difficulty Walk fast mile: No Difficulty Low Couch: A Little Difficulty Run short : No Difficulty Run 10 min: No Difficulty Vigorous activities: Limited a Little Sharp turns: A Little Difficulty Total score: 54.7 (2.3) Total score 70.9 (2.7) Admission [ ] Discharge [ ]

27 AM-PAC and KP Network Infrastructure The AM-PAC CAT application runs on a secure web server at the Kaiser Permanente Division of Research in Oakland A user calls the application from a web browser on their local computer The AM-PAC CAT web address is accessible from any computer on the Regional Kaiser Permanente network AM-PAC data can be combined with both clinical and administrative data in the electronic health record (Health Connect) to permit comparisons of functional improvement based on patient characteristics and service utilization 27

28 Research Questions: Outcomes Monitoring Study What is the feasibility of using a computer-based assessment of functional outcomes (stroke patients) across the continuum of post-acute care? What is the variation, if any, in functional outcomes of patients across multiple sites of care in a 6-month period after an acute stroke? What is the variation, if any, in functional outcomes, relative to patient characteristics or clinical pathways, service delivery? 28

29 Study Procedure: Outcomes Monitoring System 280 patients enrolled and tracked thus far Use surrogate decision-makers or proxies, if needed Assess stroke severity using the NIH Stroke Scale Assess functional outcomes with the AM-PAC CAT assessment instrument A research assistant performs assessments on site or by phone, with professional staff or family proxies if patient was unable due to competency Study staff follow participants for 6 months; assessments at discharge from each level of care and at 6 months 29

30 AM-PAC Recovery Pathways: Stroke Change: Mobility on Two Care Paths 60 ecat Scores Any SNF no IRF Any IRF Baseline Visit 2 Visit 3 6 Mo Time Points 30

31 KP Functional Outcomes System: Specific Aims Interface the AM-PAC CAT instrument with the Kaiser Permanente clinical databases (which include demographic elements, pre-morbid and co-morbid conditions, resource utilization, and interventional data) to form the Kaiser Permanente Functional Outcome System (KP-FOS) for use in comparative effectiveness studies of post acute care after stroke. Utilize KP-FOS to analyze data across different post-acute settings, particularly IRF and SNF settings, to identify the most effective post-acute care approaches to promote functional recovery after stroke. 244 more patients to create a cohort of more than 500 (NINDS-Sponsored Challenge Grant, 2009) 31

32 Application of the Study Methods to TBI Kaiser Permanente Traumatic Brain Injury Research

33 Available Variables for TBI Research Age, gender, race/ethnicity, rural/urban residence, socioeconomic status (zip code) Primary diagnosis, other injuries, co-morbid conditions; preexisting conditions Medications Laboratory studies Imaging studies Neuropsychological evaluations Services received within the health system (rehabilitation; mental health; chronic pain management; palliative care, etc.) Some data from trauma centers outside KP through outside services data bases Return to work or school data Future opportunities: genetic profiles/kp genetics project 33

34 TBI Outcomes Research: Variation in Care Opportunities for studying differences in outcome within an integrated health system by analyzing different cohorts based on variation in care Geographic differences Different referral patterns Different treatments or care settings Differences in care over time Requires: Restricted cohorts Severity measures Functional measures Characterization of services and care delivery 34

35 Available Outcome Variables for TBI Research Outcomes Complications Re-hospitalization Late effects: dementia? Mortality (California data, so some delay in obtaining) Return to work Return to school Functional outcomes if AM-PAC or FIM data 35

36 Population Comparisons in CER/TBI Research: Examples for TBI Timing Diagnosis (primary & secondary): early vs. late Treatment: early vs. late Type of Encounter or Intervention Acute Hospital (moderate to severe) Emergency care and outpatient referral (mild) Referral to TBI specialist MD (e.g. physiatrist, neurologist, psychiatrist, neuropsychologist) Neuropsychological assessment Rehabilitation therapies: cognitive, physical, psychological Referral to the Department of Rehabilitation

37 Challenges and Caveats Complexity of patient populations Complexity of care delivery Changes in care delivery over time? Generalizability to other populations and health systems Data bases: weaknesses and constraints Research methodology: weaknesses and constraints 37

38 The Truth (My Opinion, Of Course) Health systems are better at measuring than improving - Carolyn Clancy, AHRQ Our data analyses pose more questions than we are able to answer Research is not moving quickly enough for translation of findings into changes in practice Most health systems are not integrated enough, supported with research-related EMRs, to do the kind of research that needs to be done: VA and KP are best models for this Incentives must change for us to do this work on a largescale Certain crucial data elements are not available: severity measures and functional measures 38

39 Collaborators and Sponsors: KP Division of Research University of Washington Boston University CDC NIH Clinical Center National Institute of Neurological Diseases and Stroke Kaiser Permanente Community Benefits

40 Disparities in Post-Acute Care Kaiser Foundation Rehabilitation Center M. Elizabeth Sandel, MD, Principle Investigator Hua Wang, PhD, Research Scientist Richard Delmonico, PhD, Chief of Neuropsychology Kaiser Permanente Division of Research Joseph Terdiman, MD, PhD Steven Sidney, MD, PhD Charles Quesenberry, PhD Co-Investigators: Bernadette Ford Lattimer, CDC; Jeanne Hoffman, Marcia Ciol, University of Washington, and Leighton Chan, NIH Funding: Centers for Disease Control and Prevention SIP

41 Stroke Mortality in Post-Acute Care Kaiser Foundation Rehabilitation Center Principle Investigator: Hua Wang, PhD, Research Scientist M. Elizabeth Sandel, MD, Co-investigator Michelle Camicia, MSN, Co-investigator Kaiser Permanente Division of Research Joseph Terdiman, MD, PhD Arthur Klatsky, MD Steven Sidney, MD, PhD Maryann Armstrong, MA Funding: Kaiser Permanente Community Benefits 41

42 Stroke Onset Days to Rehabilitation Kaiser Foundation Rehabilitation Center Principle Investigator: Hua Wang, PhD, Research Scientist Michelle Camicia, MSN, Co-investigator M. Elizabeth Sandel, MD, Co-investigator Kaiser Permanente Division of Research Joseph Terdiman, MD, PhD Co-investigator 42

43 Outcomes Monitoring System Kaiser Foundation Rehabilitation Center M. Elizabeth Sandel, MD, Principle Investigator Marian TeSelle, MD, Co-investigator Hua Wang, PhD, Co-Investigator Richard Delmonico, PhD, Co-Investigator Michelle Camicia, MSN, CRRN, Co-Investigator Jed Appelman, PhD, Research Manager Kaiser Permanente Division of Research Joseph Terdiman, MD, PhD, Co-Investigator National Institutes of Health Clinical Center Leighton Chan, MD, MPH, Co-Investigator Elizabeth Rasch, PT, PhD, Co-Investigator Boston University Alan Jette, PT, PhD, Co-Investigator Funding: National Institutes of Health Clinical Center 43

44 KP Functional Outcomes System Kaiser Foundation Rehabilitation Center M. Elizabeth Sandel, MD, Principle Investigator Marian TeSelle, MD Co-Investigator Hua Wang, PhD, Co-Investigator Richard Delmonico, PhD, Co-Investigator Michelle Camicia, MSN, CRRN, Co-Investigator Jed Appelman, PhD, Research Manager Kaiser Permanente Division of Research Joseph Terdiman, MD, PhD Co-Investigator Boston University Alan Jette, PT, PhD Co-Principal Investigator NIH Clinical Center Leighton Chan, MD, MPH Elizabeth Rasch, PT, PhD National Institute of Neurological Disease and Stroke: Grant #1RC1NS

45 TBI and Dementia Kaiser Foundation Rehabilitation Center M. Elizabeth Sandel, MD Principal Investigator Hua Wang, MD, PhD Co-investigator Richard Delmonico PhD Co-investigator Jed Appelman, PhD Research Manager Kaiser Permanente Division of Research Rachel Whitmer, PhD Co-investigator Maryanne Armstrong, MA, Co-investigator Funding: Kaiser Permanente Community Benefits

46 Neurotrauma Conference/Las Vegas, 2010 THANK YOU! KFRC 46