Childhood Asthma Quality Care: Estimating Impact

1
Childhood Asthma Quality Care Estimating Impact
Implications

• The State of Childhood Asthma Future
  Directions
• Strategies for Implementing Best Practices
• December 13, 2006
• Soeren Mattke

2
Background

• Asthma is an inherently treatable disease, but
  actual treatment falls far short of recommended
  care
• The complexity of treating asthma requires
  innovative ways of delivering and organizing care
• Fundamental change requires actionable data on
  costs and benefits of improving asthma care

3
Technical Approach

• Systematic review of published and unpublished
  literature on gaps in asthma care and their
  impact
• Extraction of information into a database using a
  comprehensive coding tool
• Summative analysis of identified studies
• Focused analysis of three common endpoints
• Treatment rates with inhaled steroids
• Hospital admission rates
• Emergency Room visit rates

4
Items for Abstraction
Screening by title and abstract
Screening by reading articles
5
Findings on Gaps in Quality

• Most studies focus on pharmacologic treatment
• Use of anti-inflammatory medications 72 studies
  
• Median utilization rate was 32 and rates were
  typically well below 100
• Lower utilization rates found among certain
  demographic groups (publicly insured, non-white)
• Use of bronchodilators 29 studies
• Lower utilization rates reported (10-25
  theophylline 5-15 long-acting beta-agonists),
  reflecting their decreasing role
• Use of rescue drugs 56 studies
• Wide range of estimates but few studies commented
  on adequacy of use

6
Findings on Endpoints of Care

• Hospital admissions 83 studies
• Hospitalization rates generally were below 10,
  although some were as high as 35 in a study at
  an urban hospital.
• ER visits 78 studies
• Percentage visiting the ER was usually below 40,
  although some studies selected subjects on the
  basis of having gone to the ER.
• Studies where the majority of the subjects were
  white had higher ER utilization rates than
  studies where at least 40 of the subjects were
  African American.
• Missed school days 22 studies
• Many students missed days due to asthma.
• One study found that 35 of children missed at
  least one day in the past four weeks.
• Symptom free days or symptom days 5 studies
• One found fewer than 3 symptom free days on
  average in the past 2 weeks.
• Few studies explicitly attribute events to gaps
  in quality

7
Findings on Cost of Closing Gaps in Care

• 45 studies examined an asthma disease management
  program (ADMP) or a similar type of intervention.
• However, only 6 studies discussed the cost of the
  intervention or the savings it achieved.
• Insufficient evidence to estimate the
  cost-benefit ratio or return on investment for
  better asthma care

8
Analyzing Selected EndpointsRationale

• Abstraction of studies showed wide variation of
  estimates for quality of care and endpoints of
  care
• Studies used very different definitions to
  measure quality of care and endpoints
• To check whether we can explain part of the
  variance, we attempted to
• Rebase common endpoints onto a standard metric
• Account for differences in the study populations
• Disease severity
• Age composition
• Racial/ethnic composition
• Data source

9
Analyzing Selected EndpointsMetrics

• We identified three endpoints that are common
  enough for analysis
• Inhaled corticosteroid (ICS) treatment
• ER visits
• Hospital admissions
• We identified the subset of studies that would
  provide data on one of the three endpoints
• Endpoints were rebased onto a common metrics
• ICS treatment rates (n30)
• Number of ER visits per 100 patient years (n36)
• Number of hospital admissions per 100 patient
  years (n36)

10
Analyzing Selected EndpointsExplanatory
Variables

• Disease severity
• We classified as average severity study
  populations that were identified based on the
  HEDIS algorithm or equivalent criteria. Low and
  high imply less and more stringent criteria,
  respectively.
• Several studies used non-utilization based
  criteria (NHBLI, ATS), but rarely provided a
  breakdown of results by severity
• Age composition children, non-adults, adults,
  all
• Racial/ethical composition no info, majority
  white, gt40 Black, gt25 Hispanic/lt40 Black
• Data source private insurance claims, public
  insurance claims, medical records, self-report

11
We Found a Wide Range of ICS Utilization Rates

• The studies showed considerable variability in
  ICS utilization rates
• However, over half of the studies reported less
  than rates of less than 40, suggesting many
  patients receive inadequate care.

12
Adjusting for Asthma Severity did not Eliminate
the Variability in the ICS Utilization Rates

• ICS utilization rates appear to increase with
  asthma severity
• However, a wide range of estimates still exists
  within each severity category
• Numerous studies report low utilization rates
  even for patients diagnosed using HEDIS criteria
  (or something similar)

13
We Also Found Considerable Variability in ER and
Hospital Admissions

• ER and hospital admission rates varied widely
  across studies.

14
Higher ER Utilization Rates (per 100 Patient
Years) Were Found in Studies with Higher Severity

• The estimated ER utilization rates tended to be
  larger in populations with more severe asthma.
• But as with ICS utilization, there is a wide
  range of estimates within each of our severity
  categories.

15
Similar Patterns Were Found for Hospital
Admission Rates (per 100 Patient Years)

• Hospital admission rates tended to be higher in
  populations with more severe asthma.
• But the estimates vary widely within severity
  category.

16
We Also Looked at Other Characteristics of the
Study Samples

• In addition to asthma severity, we classified
  studies based on
• Race and ethnicity
• Age
• Source of data (e.g., private insurer claims)
• Some patterns were clear. For instance,
  non-whites receive less adequate care than whites
• Median estimate of ICS utilization rate
• 51 if most subjects were white
• 32 if at least 40 of subjects were black

17
Studies That Used Self-Reported Data Had Higher
Utilization Rates
ICS Utilization

• ER and hospital admission rates were also
  considerably higher in studies that used
  self-reported data.
• However, estimates vary widely among studies that
  use the same type of data.

18
Multivariate Analyses did Little to Reduce
Variability Across Studies

• We made two-way tabulations based on asthma
  severity and the racial makeup of the sample.
• There was still considerable variability
• In the 4 studies where the sample was at least
  50 white and that used the HEDIS criteria, ICS
  utilization ranged from 27 to 94
• Multivariate regression analyses did not reveal
  significant correlations between sample
  characteristics and utilization measures.

19
Summary

• Overall body of literature is large
• Types of papers are very heterogeneous, typically
  focusing on one piece of the puzzle
• Studies use widely different criteria for patient
  identification, selection and stratification as
  well as different endpoints and different
  operationalizations of endpoints
• Some patterns emerge
• Whites appear to receive better care than
  non-whites
• ER and hospitalizations more common in
  populations with more severe asthma
• Utilization rates are higher in studies based on
  self-reported data
• However, there is wide variation in results which
  cannot be explained by obvious differences in
  casemix or source of data.

20
Why is this so difficult?

• Number of studies is too low for a true
  multivariate analysis
• Accounting for casemix is very crude
• Conversion to common metrics reduces information
• Least common denominator approach
• Critical information is often not reported
• There is also substantial true variation by
• Insurance status and other determinants of access
  
• Local practice patterns

21
Recommendations

• Better estimates for the gaps in asthma care as
  well as the cost and the benefits of closing it
  are needed for various populations
• Economic research into asthma care should be
  based on consistent and standardized definitions
  to facilitate integration of knowledge

22
How can we get there?

• Generating better estimates
• Building economic evaluation into projects
• Using routinely available data (e.g., claims)
• Standardizing definitions
• Consensus workshop of researchers, advocacy
  organizations and funders