Transcript Pharmaceutical Research Databases and Consulting Capabilities

Do You Know Who Your High-Cost Asthma Patients Will
Be Next Year?
Identifying High-Cost And High-Use Asthma Patients And
Associated Treatment Costs
Ronald J. Ozminkowski, Ph.D.
Onur Baser, Ph.D.
John Azzolini, MBA, MPH
Kathe Fox, Ph.D.
October 14, 2003
Overview
• Background and Expectations About Quality / Cost Relationships
• Using MarketScan® Research Databases to Learn About These
Relationships
• Finding Asthma Patients in Claims Data
• Recording Drug Use With Quality of Care in Mind
• Estimating the Consequences of Drug Use
• Simulating Experience in a Typical Health Plan
• Simulation Results: What Do They all Mean?
• Questions
2
Copyright 2003 Thomson Medstat
Background And Expectations About Quality And Cost
• Asthma is a chronic condition characterized by inflammation of the
airways and constricted bronchial tubes.
• Symptoms range from mild (e.g. cough) to life-threatening inability to
breathe.
• About 6% of the U.S. population has asthma. Studies suggest that::
– More than 5,000 people die from asthma attacks each year.
– 3rd most prevalent chronic condition among children.
– Prevalence and mortality rates are increasing
• Studies also note that utilization and disability from asthma are high:
– Asthma accounts for over 450,000 hospitalizations,
– 13.7 million ambulatory care visits, and
– 100 million restricted activity days annually.
• But over 40% of this use is preventable (Weiss et al., 1992)
3
Copyright 2003 Thomson Medstat
Background And Expectations About Quality And Cost
• The premise behind disease management programs for asthma:
Better treatment will reduce airway inflammation, increase patients’
health status and save money.
• Many studies show that treatment with drugs to reduce
inflammation substantially improves patient outcomes
(Ozminkowski, et al. 2000).
• But other studies also show that many asthma patients do not
receive drugs to reduce airway inflammation.
• Results shown here will corroborate this.
• Results illustrate that the movement toward better care may be
happening too late, costing (rather than saving) money in the
short-run.
4
Copyright 2003 Thomson Medstat
MarketScan® Research Databases Are Key Sources Of
Data For These Analyses
MarketScan Includes:
• Large multi-source databases of privately insured claim and
encounter data
– These sources ensure that MarketScan reflects the complexity of
the real world of healthcare delivery
– Databases represent true continuum of care, including carve-outs
• Longitudinal data
– Databases support person-level analyses, with multiple years of
follow up
• The basis of more than 80 peer-reviewed journal articles in the
past six years
5
Copyright 2003 Thomson Medstat
MarketScan Features/Enhancements
• Fully adjudicated claims data from over 50 large employers with over
150 health plans
• Consistent data definitions across all employers and plans
• Medical claims can be linked to benefit plan design, enrollment, drug,
and absenteeism data
– Inpatient, outpatient, drug and enrollment data are used here.
• Patient identifiers are consistent across years, allowing us to track
asthma patients over time.
• Linked drug files contain National Drug Code, therapeutic class,
copayment, total payment, etc., to help model asthma drug use.
• Clinical classification systems such as Major Diagnostic Category,
Diagnosis Related Group, DCGs, and Disease Staging can be applied.
– DCGs were used for this analysis to control for casemix
6
Copyright 2003 Thomson Medstat
MarketScan Commercial Claims And Encounters
(CC&E) Data Were Used
• Inpatient and outpatient medical claims were linked to drug and
enrollment
• Asthma patients were found in 2000 and followed into 2001 for
these analyses.
– All were continuously enrolled for both years.
• FFS plans were represented
–
–
–
–
–
7
Exclusive provider organizations
Preferred provider organizations
Point of service plans
Indemnity plans
Capitated plans were excluded because of incomplete payment
data.
Copyright 2003 Thomson Medstat
Finding Asthma Patients
• Medstat’s Episode Grouper (MEG) software was used to find
asthma patients in the year 2000.
• MEG used ICD-9 diagnosis code 493 to find over 50,000 asthma
patients.
• Patients who were pregnant or had COPD were excluded.
• MEG also differentiated between asthma and episodes of other
acute and chronic conditions, but anyone with at least one
asthma episode was selected for analysis.
8
Copyright 2003 Thomson Medstat
Finding Asthma Patients (cont’d.)
• MEG applied Medstat’s Disease Staging software to distinguish
between asthma patients with mild (n = 44,861) versus moderate
or severe (n = 5,614) problems.
– Severe asthma requires evidence of dyspnea and wheezing,
along with hypoxemia or respiratory failure.
• Claims for asthma patients were analyses with the Diagnostic
Cost Grouper (DCGs).
– DCGs produced indicators for having 184 different conditions
affecting all body systems.
– A single DCG relative risk score was also generated to measure
the relative costliness of each patient’s conditions, compared to
the average patient in the sample.
– These measures allowed us to control for comorbidities.
9
Copyright 2003 Thomson Medstat
Recording Drug Use With Quality Of Care In Mind
• Clinical practice guidelines stress the need to control airway
inflammation, because that is the root cause of most problems for
asthma patients.
• Several drugs can be used to control for inflammation:
–
–
–
–
Inhaled steroids
Cromolyn
Nedocromil
Leukotrine modifiers
• Other drugs are meant to provide quick relief of symptoms of
inflammation (e.g., cough, breathing difficulties):
– Short-acting beta agonists
– Bronchodilators
– Anticolinergics
• We produced indicators for the types of drug used in the year 2000.
10
Copyright 2003 Thomson Medstat
Estimating Relationships Between Drug Use And “Bad”
Outcomes
• Multiple regression analyses were used to estimate the likelihood that
bad events would occur in 2001, based upon asthma drug use in 2000.
• This timing was meant to facilitate the understanding of causality.
– Actions in 2000 may have consequences in 2001.
• Bad events included:
– Having really high medical expenditures (> 80th percentile)
– Having one or more ER visits for asthma
– Having one or more hospital admissions for asthma
• Logistic regressions were used to analyze binary (yes or no) outcomes.
• Log-transformed ordinary least squares analysis with a smearing
estimate was used to estimate relationships between drug use in 2000
and medical expenditures in 2001, for those with really high
expenditures.
11
Copyright 2003 Thomson Medstat
Simulating Experience In A Typical Health Plan
• Results from regressions can be hard to interpret on their face.
• So we transformed the results to show the implications of drug
choice for a typical health plan with 100,000 beneficiaries.
• For 2001, we reported:
–
–
–
–
Expected no. of patients with really high expenditures
Average expenditures when really high
No. of patients expected to have any hospitalizations
No. of patients expected to have any ER visits
• These were reported per 100,000 enrollees, according to type of
asthma drugs taken.
• Separate reports were generated for mild vs. moderate / severe
asthma cases, assuming a 6% overall prevalence of asthma in
the plan.
12
Copyright 2003 Thomson Medstat
Simulation Results (see Table 1 in handout)
• In a plan with 100,000 enrollees:
– About 6,000 (6%) would have asthma
– Most of these (about 5,460, or 91%) would have mild asthma
• Among the 540 moderate / severe asthmatic patients:
– The number who have really high expenditures (> 80th percentile) would be
about 60% to 120% higher for those who take inflammation controllers or
combination therapy, compared to those who do not.
– Expenditures would be about 23% to 68% higher for those who take
inflammation controllers or combination therapy, compared to those who
do not
– The number who would be hospitalized would be about 1.6 to 2.0 times as
high among those who take inflammation controllers or combination
therapy.
– And the number who have any ER visits would be about 21% to 35%
higher for those who take inflammation controllers or combination therapy.
• This is the opposite of what we expected!!!
13
Copyright 2003 Thomson Medstat
Simulation Results (cont’d.)
• Among the 5,460 mild asthmatic patients:
– The number who have really high expenditures (> 80th percentile)
would be about 56% to 80% higher for those who take
inflammation controllers or combination therapy, compared to
those who do not.
– Expenditures would be about 8% to 31% higher for those who
take inflammation controllers or combination therapy, compared to
those who do not.
– The number who would be hospitalized would be about 1.3 to 2.2
times as high among those who take inflammation controllers or
combination therapy.
– And the number who have any ER visits would be about 12% to
17% higher for those who take inflammation controllers or
combination therapy.
• Not as bad, but still the opposite of what we expected
14
Copyright 2003 Thomson Medstat
Interpreting the Results
• Guidelines say that inflammation controllers should be used,
• But these analyses said the people who used them were more
expensive and more likely to have bad things happen to them.
• Does this mean disease management would cost you money?
• Maybe not!
– We cannot differentiate between those who participated in formal
disease management programs and those who did not.
– We focused on all patients, not just new asthma patients, so we do
not know if some people used inflammation controllers before
2000 and then stopped using them before we sampled them.
– Maybe people are getting these drugs too late, after the
precursors to bad events are already in motion and unstoppable.
15
Copyright 2003 Thomson Medstat
Interpreting the Results (cont’d.)
• In this sample, almost everyone had some drug therapy:
– 95% of mild patients received drug therapy
– 99% of moderate or severe patients received drug therapy of
some kind.
• But fewer received inflammation controllers
– 65% of the mild asthmatic patients
– 74% of the moderate/severe patients
• So guidelines would suggest that too few people used these
drugs in the year 2000.
• We think the timing of this use may have been inappropriate as
well.
• They did not seem to be used as first-line therapy, though they
are promoted as such and experts intend them to be.
16
Copyright 2003 Thomson Medstat
What Should Patients Do?
• Learn about their disease. Own up to the fact they have asthma
and take an active role in managing it.
–
–
–
–
Learn how to interpret symptoms and when to seek help.
Learn how to use inhalers.
Use other drugs as prescribed.
Ask doctors about other/better ways of treatment from time to
time.
– Ask for, get, and follow a written treatment plan for maintenance
drugs and rescue therapies.
– Investigate access to on-line, phone, or walk-in help from
providers when questions arise or when trouble happens.
17
Copyright 2003 Thomson Medstat
What Should The Health Plan Do?
• Health plans should work with doctors and pharmacists to Investigate
the use of inflammation controllers.
– Who is getting them?
– Are guidelines being followed?
– Are they being prescribed too late?
• Then monitor those who take these drugs and those who do not.
• Generate a list of acceptable reasons for deviating from guidelines
– For example, long-term steroid use may be inappropriate for young
children.
• But push inflammation control as a first-line of defense for most cases.
• And continue to monitor drug use, health status, and expenditures to
evaluate the impact of appropriate drug use.
• Too little too late may cost you money and negatively impact the health
of your patients.
– Adhere now, or pay later!
18
Copyright 2003 Thomson Medstat
References
Ozminkowski RJ, Wang S, Marder WD, Azzolini J, Schutt D.
Cost implications for the use of inhaled anti-inflammatory
medications in the treatment of asthma. Pharmacoeconomics
2000;18:253-264.
Weiss K, Gergen P, Hodgson T. An economic evaluation of
asthma in the United States. N Engl J Med 1992;326:862-866.
19
Copyright 2003 Thomson Medstat