Introduction • Question: How to get clinicians to change

Factors Affecting Compliance with Diabetes
Hypertension Guidelines
Julie C. Lowery, PhD, MHSA
Sarah L. Krein, PhD, RN
Lee A Green, MD, MPH
Leon Wyszewianski, PhD
Hyungjin Myra Kim, ScD
Christine P. Kowalski, MPH
• Question: How to get clinicians to change
their practices and adopt practice
guidelines for treating hypertension in
patients with diabetes?
VA HSR&D Center of Excellence
University of Michigan Departments of Health Management
& Policy, Family Medicine
Ann Arbor, MI
Ann Arbor, Michigan
• Educational strategies are most common,
but seldom result in lasting practice
• Other strategies work some of the time,
but none works all of the time.
• Clinicians are classified into four categories
based on their usual responses to new research
findings about the effectiveness of clinical
practices: seeker, receptive, traditionalist,
• Classification is based on scores from 3
subscales: evidence, practicality, conformity.
• New perspective by Wyszewianski and
Green provides a guide for selecting the
most effective change strategies for a given
group of physicians:
Wyszewianski L, Green LA. Strategies for
changing clinicians’ practice patterns: A new
perspective. J Fam Pract 2000;49:461-4.
• A valid physician classification instrument
could be used by clinicians and managers,
both within and outside VHA, to tailor the
design of their research translation or
guideline implementation efforts to the
types of physicians in their organizations,
thereby improving the effectiveness of
their efforts.
• To evaluate the construct validity and
reliability of the physician classification
Primary Hypothesis
• Compliance with medication guidelines for
diabetes patients with hypertension varies
by physician category and guideline
implementation strategy.
• Cross-sectional, observational design.
• Logistic regression, clustering within provider.
• Primary and secondary data collection.
• DV: Adherence to medication guidelines
• IVs: Site implementation strategies, physician
category (defined two different ways).
Data Collection: Phase I
• IV: Site implementation strategies.
Semi-structured telephone interviews were
conducted with 2 clinical representatives at 43
participating VA medical centers to determine
what strategies were implemented for meeting
diabetes hypertension guidelines in the time
period from 1999-2001.
Data Collection: Phase II
• IV: Physician categories.
All primary care physicians (PCPs) in the
participating VAMCs were sent a one-page
questionnaire (the physician classification
instrument) regarding their responses to research
findings about the efficacy of specific clinical
practices. [Instrument.]
Data Collection: Phase III
• DV: Adherence to medication guidelines
Defined as: % of each participating
physician’s patients with diabetes and HTN
who were on HTN meds at time of elevated
BP reading, or who had ↑ in dosage or Δ in
med class in 6 months following reading.
Data Collection: Phase III
• Diabetes = (1) had filled a prescription for diabetes medications
or blood glucose monitoring supplies; or, (2) had 1 inpatient or 2
outpatient encounters with a diabetes related ICD-9 code (250.x,
357.2, 362.0-362.1, 366.41) in fiscal year (FY) 1999.
• HTN = BP > 140/90 mmHg.
• HTN meds: ace inhibitors, beta blockers, calcium channel
blockers, alpha blockers, angiotensin II inhibitors, diuretics.
• Data sources: VA secondary data sets with data on
vitals, medications, diagnoses.
• Time frame: October 1998 – March 2000.
Data Collection: Phase III
Results: Phase I
• Data sources: VA secondary data sets with
data on vitals, medications, diagnoses.
• All sites used some type of educational approach
to implement the guidelines (written,
presentation, or conference).
• Time frame: October 1998 – March 2000.
• Patient data were matched to each
participating PCP.
• Over 90% of sites also provided group or
individual feedback on physician performance
on the guidelines, and over 75% implemented
some type of reminder system.
• Minority of sites used monetary incentives,
penalties, or barrier reduction.
Results: Phase II
Results: Phase III
• Of 745 questionnaires distributed to primary
care physicians, 307 were returned (response
rate of 41.2%).
• 174 pragmatists (59.8%)
• Of 307 questionnaires returned, 16 had missing
data, leaving a total of 291 useable
• 1 traditionalist (0.3%).
• 80 receptives (27.5%).
• 36 seekers (12.4%).
• Factor analysis confirmed the 3-factor
psychometric scaling used previously (2
questions dropped).
Results: Phase III
• The total number of diabetes patients in the 42
participating sites was 208,653 in 1999.
• Patients in the diabetes cohort were assigned to
participating PCPs if more than 50% of a
patient’s outpatient medical clinic visits were to
a participating PCP.
Results: Phase III
• 1st method of measuring interaction between
intervention strategy and physician category
(concordance scores) Æ no association with
guideline adherence.
• Final dataset: 1174 diabetes patients had BP
data, had HTN, and could be matched to 163 of
our participating PCPs.
Results: Phase III
• 2nd method of measuring interaction between
intervention strategy and physician category:
• Interventions were coded as the number of
educational interventions, barrier removal
interventions, and motivational interventions (3
• Physician characteristics coded as scale scores (3
Results: Phase III
Testing full model with all 2-way interactions between
interventions and physician scale scores:
• The only interaction that approached significance was
conformity with barrier removal (p = 0.07). Barrier
reduction was associated with improved guideline
concordance for the least conforming physicians, but
not for the conforming physicians. [Figure]
• Significant positive association of barrier removal with
guideline concordance (p = 0.03).
Results: Phase III
Testing without interactions:
• Only conformity scale was significantly
associated with guideline concordant-care.
Lower conformity was associated with better
• No association between guideline interventions
and guideline concordant care.
Results: Reliability
• One-year test-retest results. Of the 291 participating
providers with useable surveys, 263 (90%) completed
follow-up surveys one year later. The correlations for
the three subscales were as follows:
– Evidence: .75
– Practicality: .68
– Non-comformity: .75
• These findings suggest that physician scores on the
three subscales remain relatively stable over time,
indicating that the concept of physician response to new
information is more of a trait than a state.
• Main conclusions:
– Non-conformity is associated with better guideline
– Barrier reduction is associated with better guideline
– As conformity increases, the impact of barrier
reduction decreases.
• Guideline implementation strategies that were designed
to reduce, or at least not increase, physician time
demands and task complexity were the only ones that
improved guideline adherence—particularly for
physicians low on the conformity scale. In other words,
the more physicians were willing to practice differently
from the local norm, the more they took advantage of
system changes to change their own practices.
• Education may have been necessary, but it was
clearly not sufficient; all sites included education
in their mix of strategies, but those doing a great
deal of it saw no more effect than those doing
the minimum.
• Incentives had no discernible effect.
• Primary hypothesis was not valid—no
association between physician type/intervention
interaction (measured by concordance scores)
and guideline adherence.
• Possible explanation: time constraints of current
delivery environment?
• Is an instrument for measuring physician type
• Limitation: Small sample size.
• Focus of interventions should be at the system or
organizational level, rather than the provider.
• Results consistent with other studies Æ quality
improvement efforts should focus on addressing
facility-level performance variations, because of
the small amount of variation in performance
found at the provider level.