Predictors of Preventive Services

1
Predictors of Preventive Services Use Among
Medicare Beneficiaries

• Ronald J. Ozminkowski, Ph.D.
• Ron Z. Goetzel, Ph,D.
• David Shechter, Ph.D.
• David C. Stapleton, Ph.D.
• Onur Baser, Ph.D.
• Pauline J. Lapin, M.H.S.
• Thomson Medstat (RJO, RZG, DS,OB)
• Cornell University (RJO, RZG, DCS)
• Centers for Medicare and Medicaid Services (PJL)

2
Background

• Clinical preventive services have been shown to
  prevent disease and promote early detection and
  treatment.
• Despite Medicare coverage, use of clinical
  preventive services has been less than optimal.
• Little is known about the factors that determine
  the use of preventive services among Medicare
  beneficiaries.
• Information on factors that predict the use of
  preventive services use among Medicare
  beneficiaries could inform policymakers of
  actions they might consider to promote
  appropriate use.

3
Background

• The Medicare Prescription Drug, Improvement, and
  Modernization Act of 2003 provides Medicare
  reimbursement for an initial preventive physical
  exam (a.k.a the Welcome to Medicare visit)
  designed to review the health status of new
  Medicare beneficiaries.
• It is expected that this visit will yield an
  increase in the use of preventive services that
  Medicare pays for, including
• Pneumococcal, influenza, hepatitis B vaccinations
• Screening mammography
• Screening pap smear and pelvic exam
• Colorectal cancer screening tests
• Prostate cancer screening tests
• Glaucoma screening
• Bone mass measurements

4
Hypotheses and Data Source

• We expect the number of preventive services used
  to be influenced by
• Demographic and socioeconomic factors
• Health plan type
• Health status
• Underlying health risks
• Ability to take care of daily needs and
• Motivation to take care of oneself.
• The 2001 Medicare Current Beneficiary Survey
  (MCBS) supplied data to test these hypotheses.
• 11,158 respondents with non-missing data

5
Variables Used in Analyses of Predictors of
Preventive Services Use

• The 2001 MCBS provided information on more
  variables expected to influence preventive
  services use than in other studies.
• Demographic and socioeconomic factors
• Gender
• Race
• Marital status
• Employment status
• Income
• Education
• Number of living children
• Living alone or not
• Place of residence (detached home, retirement
  community)
• Availability of personal care services
• Health plan type
• HMO or not
• Dually enrolled in Medicare and Medicaid or not

6
Variables Used in Analyses of Predictors of
Preventive Services Use (Contd.)

• Health status measures
• Having history of one or more of these
  conditions
• Hardening of the arteries
• Hypertension
• Myocardial infarction
• Angina or coronary heart disease
• Other heart problems
• Stroke
• Diabetes
• Rheumatoid arthritis
• Psychiatric / mental health problems
• Osteoporosis
• Broken hip
• Emphysema / asthma / COPD
• Complete or partial paralysis
• Cancer related to screening test of interest
• Cancer not related to screening test of interest
• Depression
• Loss of general interest in life in last 12 months

7
Variables Used in Analyses of Screening Behavior
(Contd.)

• Health status measures (Contd.)
• General health status (excellent, very good,
  good, fair or poor)
• Health risks
• Smoking status (current smoker, former smoker)
• Overweight
• Obese
• Drinking habits (heavy drinker vs. not)
• Gets at least some weekly exercise
• Does moderate, vigorous, or muscle-building
  exercise at least once / week
• Motivation
• Patient activation scale (available for 8,590
  respondents, so used in some analyses only)

8
Variables Used in Analyses of Predictors of
Preventive Services Use (Contd.)

• Ability to carry out daily activities, related
  to
• Use of telephone
• Housework (light and heavy studied separately)
• Paying bills
• Preparing meals
• Shopping
• (Analyses account for ability and desire to do
  these)

9
Statistical Methods Descriptive Analyses

• Descriptive analyses showed characteristics of
  the sample members related to these variables.
• Descriptive analyses also showed relationships
  between these variables and the probability of
  having a
• Low number of services (1 4)
• Medium number of services (5 or 6)
• High number of services (7)
• (in the past 12 months)

10
Statistical Methods Multivariate Analyses

• Multinomial logistic regression analyses showed
  impact of each variable on the probability of
  having low, medium, or high numbers of preventive
  services, controlling for the impact of all other
  variables.
• Other analyses (available upon request),
  included
• Poisson count model analyses showing the marginal
  impact of each variable on the number of services
  done in the previous 12 months (ranging from 0
  7 possible services).
• Logistic regression analyses showing the impact
  of each variable on the probability of having
  each type of service done in the previous 12
  months.
• All analyses were conducted in STATA and adjusted
  for the complex sampling process used for the
  MCBS.
• Results are nationally representative.

11
Results Characteristics of the Full Sample (n
11,158)

• Demographics and Socioeconomic status
• Female 58.4
• African American Race 7.7
• Never married 3.2
• Currently employed 12.1
• Low income (0 to 25k) 59.2
• Had no living children 9.0
• Some college education or graduated college
  69.3
• Lives alone 32.3
• Live in single family detached home 70.9
• Plan Type
• HMO members 20.8
• Dually enrolled in Medicare and Medicaid 9.4

12
Health Status of Respondents Heart Related
Conditions
13
Health Status of Respondents Depression Other
MH Problems
14
Health Status of Respondents Musculoskeletal
Problems
15
Health Status of Respondents Cancer, COPD,
Asthma, Emphysema
16
Perceived Health Status of Respondents
17
Percent Who Received Each Type of Preventive
Service in Last 12 Months
18
Results Factors Associated With Having a High
Number of Preventive Services in Full Sample

• The adjusted probability of having a high number
  of services (all 7) was significantly greater
  for
• Those with children (14.1) vs. those with no
  children (13.3)
• HMO members (14.7) and FFS members (14.4) vs.
  those dually enrolled in Medicare and Medicaid
  (10.3)
• Those with the following conditions
• Hardening of the arteries (17.4) vs. not (13.7)
• Hypertension (14.7) vs. not (13.0)
• Other heart problems (15.0) vs. not (13.8)
• Diabetes (16.9) vs. no diabetes (13.4)
• Osteoporosis (17.4) vs. no osteoporosis (13.4)
• Emphysema, asthma, or COPD (16.1) vs. none
  (13.7)
• Cancer related to the test of interest (17.3)
  vs. not (13.5)
• Those who participated in at least some weekly
  exercise (14.4) vs. not (13.4)
• Those who were more motivated to care for
  themselves (based on analysis of subsample who
  completed Patient Activation Scale questions)

19
Results Factors Associated With Having a Low
Number of Preventive Services in Full Sample

• The adjusted probability of having a low number
  of preventive services (1 4) was significantly
  greater for those who were
• Females (42.3) vs. males (34.2)
• African Americans (43.4) vs. Other races (38.5)
• Currently employed (44.7) vs. not (38.2)
• Lower income 0 - 25k (42.8) vs. high income
  (gt 50k 28.6)
• Dually enrolled in Medicare and Medicaid (42.2)
  vs. FFS (39.9) and HMO (35.0)
• With excellent general health status (45.0) vs.
  fair / poor (37.0)
• Had a broken hip (44.3) vs. not (38.7)
• Heavy drinkers (50.3) vs. not (38.8)
• Completely or partially paralyzed (43.1) vs. not
  (38.7)
• Had difficulty shopping (44.1) vs. not (38.4)
• Normally did not do shopping (44.1) vs. shoppers
  (38.1)
• Had difficulty using a telephone (40.8) vs. not
  (38.8)

20
Implications

• Having a chronic condition was often associated
  with having more preventive services.
• Statistical significance may be due in part to
  large sample size.
• Outreach to promote the appropriate use of
  preventive services should be targeted to
  segments of the population with lower use,
  including
• Females
• African Americans
• Currently employed (cannot afford to stop
  working?)
• Lower income
• Dually enrolled in Medicare and Medicaid
• Had a broken hip
• Heavy drinkers
• Special arrangements may be need for some of
  these people
• Completely or partially paralyzed
• Had difficulty shopping or who normally did not
  do shopping
• Had difficulty using a telephone

21
Implications (Contd.)

• The new Medicare drug law provides Medicare
  reimbursement for an initial preventive physical
  exam designed to review the health status of new
  Medicare beneficiaries and promote the use of
  preventive services.
• It will be important to evaluate the impact of
  this visit on the use of preventive services
  among new beneficiaries.
• It is unlikely that a one-time visit will be
  sufficient to motive people to engage in these
  practices over time.
• Future research should identify, test and
  evaluate interventions targeting segments of the
  population where use is low, and methods to
  sustain appropriate use over time.