2. Agenda
• Review CER research questions and
hypotheses
– Specific aims of the grant
– Conceptual model
– Refining hypotheses
– Sources of data
• Outline for systematic identification of data
domains and elements
3. Agenda
• Review CER research questions and
hypotheses
– Specific aims of the grant
– Conceptual model
– Refining hypotheses
– Sources of data
• Outline for systematic identification of data
domains and elements
4. Specific Aims of the Grant
• Specific Aim Related to CER (Aim 3): Develop
and enhance four sentinel cohort pairs of
patients with asthma (pediatric and adult),
hypertension, and hypercholesterolemia
distinguished by their care delivery
characteristics which can support
comparative effectiveness research.
5. Specific Aims of the Grant
• Overall Goals:
• Demonstrate the capability of the SAFTINet data
system to collect and accurately link relevant and
valid patient-level information necessary for
comparing the effectiveness of different delivery
system strategies
• Lay the groundwork (cohort identification,
outcomes measurement, sample size estimates,
etc.) to conduct prospective observational studies
and clinical trials
6. Specific Aims of the Grant
• The SPECIFIC SUB-AIMS for Aim 3 are:
– Specific Aim 3.1 Specify the data elements required
for optimal cohort creation.
– Specific Aim 3.2 Develop and use multivariable
models of asthma, blood pressure and cholesterol
control to identify system-level, individual care
provider-level, and patient-level factors associated
with the control of these conditions.
– Specific Aim 3.3 Enhance the data set by
implementing point-of-care data collection tools for
health-related quality of life.
7. Agenda
• Review CER research questions and
hypotheses
– Specific aims of the grant
– Conceptual model
– Refining hypotheses
– Sources of data
• Outline for systematic identification of data
domains and elements
8. Conceptual Model
Relatively
Mutable
CLINICAL INERTIA
Counseling
Drug selection
Dosage selection
Concomitant meds
Follow-up
Decision support
PATIENT-CENTERED
MEDICAL HOME
Integrated Mental
Health Care
Disease-specific case
mngmnt
Access to care
Outcomes feedback
THERAPY ADHERENCE
Therapy persistence
Mental health status
Health knowledge
Perceived need for
care
Symptoms
Drug side effects
PROCESSES OF
CARE
(clinician
factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic
disease
control)
Relatively
immutable
Appointment time
Patient load
Physical facilities
Practice type
Support personnel
Generalist vs.
specialist
Age
Gender
Race/ethnicity
SES
Marital status
Religious/cultural
beliefs
Comorbidity
9. Agenda
• Review CER research questions and
hypotheses
– Specific aims of the grant
– Conceptual model
– Refining hypotheses
– Sources of data
• Outline for systematic identification of data
domains and elements
10. Refining Hypotheses
• Hypothesis from Research Design: “We
hypothesize that health care delivery system
factors, such as the patient-centered medical
home, outweigh individual care provider
factors, patient factors, and medication
effectiveness in the control of asthma, high
blood pressure and hypercholesterolemia. “
11. Example Hypotheses
• Pediatric asthma outcomes (define) are better by
some amount X (define) at health centers that
implement PCMH functions (define specific
function(s)).
• A greater proportion of adult hypertension patients
with a dx of depression are appropriately controlled
at practices that have integrated mental health
services (IMH).
• The level (intensity) of IMH services is correlated
with improved BP control in adult HTN patients
whom also have a dx of depression.
12. Refining Hypotheses
CLINICAL INERTIA
Drug selection
Dosage selection
Concomitant meds
PATIENT-CENTERED
MEDICAL HOME
Intgrtd Mental Health
Disease-specific case
management
Access to care
THERAPY ADHERENCE
Therapy persistence
Mental health status
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
Generalist vs. specialist
Age, Gender
Race/ethnicity
SES
Marital status
Comorbidity
Hypothesis: health care delivery system factors, such as the patient-centered
medical home, outweigh individual care provider factors, patient factors, and
medication effectiveness in the control of asthma, high blood pressure and
hypercholesterolemia.
13. Refining Hypotheses
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
•Primary explanatory variable: patient-centered medical home
•Dependent variables: meeting national, evidence-based guidelines
for control
•Independent variables: Factors associated with disease control,
identified from literature, research experience, and clinical
judgment
•Statistical analysis: Mixed effects models will be used to determine
factors associated with chronic disease control; the primary
explanatory variable of the PCMH clinic, and the other factors
impacting chronic disease control
14. Agenda
• Review CER research questions and
hypotheses
– Specific aims of the grant
– Conceptual model
– Refining hypotheses
– Sources of data
• Sample data dictionary
• Outline for systematic identification of data
domains and elements
15. Sources of Data
• EHR
• Medicaid claims
• Enhanced point-of-care data collection
• Organizational or practice-level survey
16. Agenda
• Review CER research questions and
hypotheses
– Specific aims of the grant
– Conceptual model
– Refining hypotheses
– Sources of data
• Sample data dictionary
• Outline for systematic identification of data
domains and elements
19. Agenda
• Review CER research questions and
hypotheses
– Specific aims of the grant
– Conceptual model
– Refining hypotheses
– Sources of data
• Sample data dictionary
• Outline for systematic identification of data
domains and elements
20. Outline for Systematic Identification of
Data Domains and Elements
• Establish:
– Hypotheses and research questions
– Cohort definition
– Outcome measures
– Primary explanatory variable
– Covariates
• Establish these in order to:
– Make a list of needed data elements for current work
– Lay the groundwork for future directions
• Document a rationale for hypotheses and
selection of measures (constructs and data
elements)
21. Outline for Systematic Identification of
Data Domains and Elements
• An example from an asthma cohort
• Purpose of example to illustrate
– Selecting hypotheses and measures
– Listing data elements
– Documentation of a rationale for hypotheses and
selection of measures (constructs and data
elements)
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
22. Outline for Systematic Identification of
Data Domains and Elements
• Establish:
– Hypotheses and research questions
– Cohort definition
– Outcome measures
– Primary explanatory variable
– Covariates
• Establish these in order to:
– Make a list of needed data elements for current work
– Lay the groundwork for future directions
• Document a rationale for hypotheses and
selection of measures (constructs and data
elements)
23. Outline for Systematic Identification of
Data Domains and Elements
•Establish hypothesis: clinical inertia is
associated with worse asthma control
CLINICAL INERTIA
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
24. Outline for Systematic Identification of
Data Domains and Elements
•Primary explanatory variable: clinical inertia
•Dependent variables: meeting evidence-based
guidelines for control
•Independent variables: Factors associated with
disease control, identified from literature,
research experience, and clinical judgment
CLINICAL INERTIA
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
26. Outline for Systematic Identification of
Data Domains and Elements
• Establish:
– Hypotheses and research questions
– Cohort definition
– Outcome measures
– Primary explanatory variable
– Covariates
• Establish these in order to:
– Make a list of needed data elements for current work
– Lay the groundwork for future directions
• Document a rationale for hypotheses and
selection of measures (constructs and data
elements)
27. Cohort Definition
• Concept: patients with persistent asthma
• Options
– Research Design’s definition
– HEDIS criteria
– Adjusted HEDIS criteria
– Enhanced data collection
– Others
28. Cohort Definition
• Research Design: “As per the 2007 …EPR-3, we will
define persistent asthma as” > 1 of the following
criteria in 12 months
– > 1 prescriptions for an asthma maintenance therapy
– > 2 asthma-related ED visits
– > 1 asthma-related hospitalization
29. Cohort Definition
• HEDIS: > 1 of the following criteria in 12 months
– > 4 asthma medication dispensing events
– > 2 asthma medication dispensing events + 4 asthma-related
outpatient visits
– > 1 asthma-related hospitalization
– > 1 asthma-related ED visit
• Adjusted HEDIS criteria: improved validity if patients meeting
criteria for >2 consecutive years (Mosen et al., 2005)
• Enhanced data:
– Patient-entered chronic severity (kiosk) to assess current impairment and
future risk (Porter et al., 2004)
– Provider-entered assessment of severity
30. Data Elements for Cohort Definition
Construct Measure Elements Values Reference Data
Source
Cohort
definition
HEDIS definition of persistent
asthma 1: > 4 asthma
medication dispensing events in
12 months
Asthma medication
dispensed (date,
medication)
y/n HEDIS
manual
Claims
data
Cohort
definition
HEDIS definition of persistent
asthma 2: > 2 asthma
medication dispensing events +
4 asthma-related outpatient
visits in 12 months
Asthma medication
dispensed (date,
medication)
Asthma-related
outpatient visits
(date, ICD-9 code)
y/n HEDIS
manual
Claims
data
Cohort
definition
HEDIS definition of persistent
asthma 3: > 1 asthma-related
hospitalization in 12 months
Asthma-related
inpatient visits (date,
ICD-9 code)
y/n HEDIS
manual
Claims
data
Cohort
definition
HEDIS definition of persistent
asthma 4: > 1 asthma-related
ED visits in 12 months
Asthma-related ED
visits (date, ICD-9
code)
y/n HEDIS
manual
Claims
data
31. Rationale for Selection of Measures
The current HEDIS measure for asthma uses administrative data
collected during 1 year to identify patients with presumed
persistent asthma and evaluates controller therapy during the
next year. The current HEDIS asthma inclusion include a
significant portion of patients with intermittent asthma;1, 2 thus,
we chose to use the methods validated by Moser et al., who
adapted the HEDIS measure to require at least 2 consecutive
years meeting qualification criteria to identify persistent
asthma.3
1 Kozyrskyj AL, Mustard CA, Becker AB. Identifying children with persistent asthma
from health care administrative records. Can Respir J. 2004;11:141-145.
2 Cabana MD, Slish KK, Nan B, Clark NM. Limits of the HEDIS criteria in determining
asthma severity for children. Pediatrics. 2004;114:1049-1055.
3 Mosen DM, Macy E, Schatz M, et al. How well do the HEDIS asthma inclusion criteria
identify persistent asthma? Am J Manag Care. 2005 Oct;11(10):650-4.
32. Outline for Systematic Identification of
Data Domains and Elements
• Review
– Hypotheses and research questions
– Cohort definition
– Outcome measures
– Primary explanatory variable
– Covariates
– Other future directions
• Review these in order to
– Make a list of needed data elements for current work
– Lay the groundwork for future directions
33. Outcome Measures
• Patient-Reported Control Measures
• Utilization Measures
• Health-Related Quality-of-Life (HRQoL)
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
34. • Asthma Control Test (ACT) (Nathan et al. 2004)
• Childhood Asthma Control Test (Liu et al.
2007)
• Asthma Control Questionnaire (Juniper et al.
1999)
• Asthma Therapy Assessment Questionnaire
(ATAQ) control index (Vollmer et al. 1999) –
mentioned in Research Design
Patient-Reported Control Measures
35. Outcome Measures: Utilization Measures
• Ratio of controller to total asthma medications—
mentioned in Research Design
– > 0.5 is suggested cut-point
– better associated with utilization (ED visits) than is
HEDIS outcome measure
– Weighted vs. not
• HEDIS outcome measure
– prescription of at least one controller medication
– found to be more of a severity indicator than
quality/control measure
• Acute hospital visits (ED, inpatient)
36. Outcome Measures: HRQoL
• Asthma-Specific Quality of Life
– Mini Asthma Quality of Life Questionnaire (Juniper et
al. 1999a)
– Asthma Quality of Life Questionnaire (Katz et al. 1999;
Marks et al. 1993)
– ITG Asthma Short Form (Bayliss et al. 2000)
– Asthma Quality of Life for Children (Juniper et al.
1996)
– Others?
• Generic Quality of Life
– SF-36 (Bousquet et al. 1994)
– SF-12 (Ware et al. 1996)
37. Data Elements for Outcome Measures
Definition
Construct Measure Elements Values Reference Data
Source
Asthma
control
Childhood Asthma Control Test 7 components 0-27
(poor
control
<19)
Liu et al.
2007
POC
measure
Asthma
control
Ratio of controller to total
asthma medications
Asthma medication
dispensed (date,
medication)
0-1
(dichot
omize
at 0.5)
HEDIS
manual
Claims
data
Asthma
control
Acute hospital resource
utilization
Asthma-related
inpatient visits (date,
ICD-9 code)
# visits Claims
data
Asthma
control
Acute hospital resource
utilization
Asthma-related ED
visits (date, ICD-9
code)
# visits Claims
data
39. Outline for Systematic Identification of
Data Domains and Elements
• Establish:
– Hypotheses and research questions
– Cohort definition
– Outcome measures
– Primary explanatory variable
– Covariates
• Establish these in order to:
– Make a list of needed data elements for current work
– Lay the groundwork for future directions
• Document a rationale for hypotheses and
selection of measures (constructs and data
elements)
40. Primary Explanatory Variable
• Clinical inertia: “the failure of clinicians to
initiate or intensify drug therapy appropriately
in a patient with uncontrolled asthma, blood
pressure or cholesterol”
CLINICAL INERTIA
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
42. Data Elements for Primary Explanatory
Variable Definition
Construct Measure Elements Values Reference Data
Source
Clinical
inertia
Childhood Asthma Control Test 7 components, date 0-27
(poor
control
<19)
Liu et al.
2007
POC
measure
Clinical
inertia
Medications Asthma medication
dispensed (date,
medication)
Claims
data
44. Outline for Systematic Identification of
Data Domains and Elements
• Establish:
– Hypotheses and research questions
– Cohort definition
– Outcome measures
– Primary explanatory variable
– Covariates
• Establish these in order to:
– Make a list of needed data elements for current work
– Lay the groundwork for future directions
• Document a rationale for hypotheses and
selection of measures (constructs and data
elements)
45. Covariates
• Processes of Care
– Clinical inertia (primary
explanatory variable)
– Medication prescription
• Structures of Care
– Practice demographics
– PCMH, IMH
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
• Patient Factors
– Demographics, access
– Co-morbidity (medical,
mental health)
– Severity of illness
– Therapy adherence
46. Covariates
• Processes of Care
– Clinical inertia (primary
explanatory variable)
– Medication prescription
• Structures of Care
– Practice demographics
– PCMH, IMH
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
• Patient Factors
– Demographics, access
– Co-morbidity (medical,
mental health)
– Severity of illness
– Therapy adherence
47. Covariates
• Medical Comorbidity:
– “Chronic medical co-morbidity will be...grouped
into 30 comorbidities as described by Elixhauser
and Quan.” (AHRQ co-morbidity measures)
– Body Mass Index (do we need other measures for
children?)
– Smoking status (also 2nd hand smoke exposure?)
PROCESSES OF
CARE
(clinician factors)
+
STRUCTURES OF
CARE
(system factors)
+ PATIENT FACTORS →
OUTCOMES
(chronic disease
control)
Comorbidity
49. Data Elements for Comorbidity
Variable Definition
Construct Measure Elements Values Reference Data
Source
Medical
co-
morbidity
HCUP comorbidity measure ICD-9 codes from
encounters and
problem list
0-27
(poor
control
<19)
Liu et al.
2007
EHR
52. Outline for Systematic Identification of
Data Domains and Elements
• Establish:
– Hypotheses and research questions
– Cohort definition
– Outcome measures
– Primary explanatory variable
– Covariates
• Establish these in order to:
– Make a list of needed data elements for current work
– Lay the groundwork for future directions
• Documenting a rationale for hypotheses and
selection of measures (constructs and data
elements)
58. Patient Centered Med Home
Standards- NCQA
1. Access and Communication
2. Patient Tracking and Registry Functions
3. Care Management
4. Patient Self‐Management Support
5. Electronic Prescribing
6. Test Tracking
7. Referral Tracking
8. Performance Reporting and Improvement
9. Advanced Electronic Communications
61. System Level Factors
• Applied differently based on
patient/family/doctor-- can we account for
this or not??
62. Considerations for Future Research
Asthma:
• Asthma epidemiology has focused on individual-
level and family risk factors.
• Less focus on social and environmental context.
• Low-income individuals more likely to be exposed
to irritants, pollutants, indoor allergens, and
psychosocial stress, which may influence asthma
morbidity.
• Future vision: enhance our cohort with data on
suspected biological and environmental
determinants of asthma disparities.
63. Considerations for Future Research
Hypertension:
• Prevalence and rate of diagnosis of hypertension
in children and adolescents are increasing, due in
part to the increasing obesity prevalence and
growing awareness of hypertension.
• Future vision: expand our cohort to include
adolescents with hypertension in an effort to
identify health care delivery strategies
appropriate for the lifespan of patients with
hypertension.
64. Considerations for Future Research
Hypercholesterolemia:
• American Academy of Pediatrics recommends
screening overweight children with a fasting
lipid profile
• Rising obesity epidemic in U.S. children
• Future vision: expand our
hypercholesterolemia cohort to include
overweight children.
Notas del editor
Here are the data elements for one measure—the HEDIS measure—for the cohort definition construct. The HEDIS measure for persistent asthma has four components, each with 1-2 component elements.
To help illustrate the processes involved in constructing cohorts and associated data elements, we will run through a hypothetical hypothesis for an asthma cohort.
Would also need to include rationale for hypothesis
The cohort definition is meant to capture patients with persistent asthma. There are several options proposed in the Research Design and in other sources for how to define a cohort of patients with persistent asthma. These are summarized here, and shown in slightly more detail on the next 2 slides.
Here are the data elements for one measure—the HEDIS measure—for the cohort definition construct. The HEDIS measure for persistent asthma has four components, each with 1-2 component elements.
Here is a sample rationale-statement for the selection of one measure: the 2-consecutive-year version of the HEDIS definition of persistent asthma
There are several options proposed in the Research Design and in other sources for how to define outcomes indicative of asthma control. These are summarized here, and shown in slightly more detail on the next 3 slides. We would need to select from these (and other) measures.
This list includes some of the pt-reported control measures; we would select one of these for implementation at the POC. We would need to provide a rationale for the selection.
There are several measures of asthma control based on healthcare resource utilization; this slide is a partial list.
HRQoL is another construct that would fall into the Outcome Measures category. We would need to determine whether we will be using a disease-specific measure or a general measure for all cohorts, and then provide rationale for the specific selection
These are data elements needed for a few of the asthma-related outcome measures
As this is only an example we will not provide rationale statements for each type of data element
Our primary explanatory variable is clinical inertia; this slide shows the definition from the Research Design.