improving care for diabetic patients
TRANSCRIPT
Improving Care for Diabetic Patients
Jim Mold, M.D., M.P.H. The University of Oklahoma
Department of Family and Preventive Medicine
Project #1
Reasons physicians give for not meeting quality of care standards for diabetic patients
Jim Mold, MD, MPHBud Oehlert, MD (OFMQ)Margaret Enright, MPH, CDE (OFMQ)
Research Question
Why don’t physicians always achieve a perfect score on diabetes quality of care audits?
Methods:Approximately 2000-2001
All diabetic patients >50 years of age followed by participating OKPRN physicians for at least one year and seen by them within the last 3 months
OFMQ chart audit to determine if DQIP targets were met
Methods A1c in past year Lipid panel in past 2 years UA for protein in past year Eye exam in past year Foot exam in past year ACEI for HTN and/or proteinuria Flu shot in past year Pneumococcal vaccine ever
Methods Audit report left on each chart with a
survey instrument requesting physician to indicate reasons for not meeting each of the unmet targets
Eight fixed choices plus “other”
Results Audits are not perfect
Auditors counted off for no microalbumin when UA showed protein or patient already on an ACEI
Auditors counted off if had flu shot early last year and late this year (>1 year)
Different reasons for different quality indicators
Results/Reasons• A1c: not indicated
• UA, micro-albumin: forgot or not indicated
• Retinal exams and foot exams: done but report/findings not in chart (documentation problem)
Results/Reasons• Flu shot: offered/declined
(documentation problem)
• Pneumovax: as for flu; “inadequate reimbursement”
• BP<130/80, A1c<9.5, LDL<130: pt. making progress; non-adherence
Summary
• Optimal audit scores are less than 100% (probably 85-90% depending upon patient population)
• Improvement will probably require several different interventions (a flow sheet is not likely to improve all indicators)
Project #2
BP Control in Diabetic Patients
Adam Cotton, MS2Jim Mold, MD, MPHCheryl Aspy, PhD
Research Question
Why do PCP’s sometimes not attempt to lower BP below 130/80 in their diabetic patients?
Assumption: There are a variety of legitimate clinical reasons for not doing so.
Methods• Consecutive diabetic patients seen by
eight participating OKPRN physicians• Clinic note reviewed by a medical student• If BP>130/80 AND physician’s note did
not mention any change in strategy, student interviewed physician (within 2 weeks of the index visit)• Structured interview• Audiotaped and transcribed
Methods
• Transcribed interviews reviewed separately by the three investigators• Coded for categories of reasons
• Categories reviewed by group and differences resolved
Results
• Clinician Factors• Patient Factors• Information/Measurement Factors
Clinician Factors• Co-management (e.g. BP co-managed by
another physician)• Competing demands (e.g. patient
presented with acute problem)• Satisfied with progress/waiting for full
effect of medicine• Should generally take 6 weeks max.
• Disagreement with ADA guidelines• Only 1 of 9 physicians
Patient Factors• Limited options (e.g. financial constraints,
multiple other meds, ESRF)• Adherence problems (e.g. cognitive
deficits, mental health problem, language/cultural barrier, denial)
• Competing agendas (e.g. different goals than clinician)
• Unfavorable risk:benefit ratio
Information/Measurement Factors• Documentation error (BP or
intervention not recorded)• Insufficient or confusing information
• Patient missed dose of meds• Lack of consistent trend• Explanation/rationalization (pain, stress,
exertion)• Home readings normal/office readings
high
Conclusions
• Many reasons for not lowering BP to target• Physician factors, patient factors,
measurement factors
• Measurement factors might be ameliorated by 24 BP monitoring
Project #3
Improving Diabetes Care Using Best Practices Research and Practice Enhancement Assistants
Jim Mold, MD, MPH Margaret Enright, MPH, CDE W. H. Oehlert, M.D. Dale Bratzler, D.O. K.D. Walkingstick, MS
Research Question• Can the quality of diabetes care be improved by
a three part intervention:• Feedback on performance with benchmarking• Instruction of clinicians in principles derived from
exemplar interviews • Practice enhancement assistants to facilitate practice
changes
• Compared to clinician feedback/benchmarking alone?
Methods• Pre- and post-intervention change with
historical comparison group that received feedback with benchmarking
• All audits performed by trained OFMQ auditors
• Duration of Study: 9 months• 1 month to identify the “best practice” principles• 4 months of pre-intervention data (June-Aug)• 4 months of post-intervention data (Oct-Jan)
Outcome Measures
• DQIP Indicators (same as for study #1)
• We also collected data on mammography (within 2 years) as a control variable
Methods (cont.)
• From existing audit data, OFMQ staff identified 5 OKPRN clinicians with exemplary performance
• 90% of records met two or more of diabetes care indicators
• Two or more exemplars for each diabetes care indicators
Methods (cont.)• Exemplars interviewed by OFMQ nurse by
phone• Interviews transcribed
• From transcripts, three researchers identified and agreed upon a set of 6 principles of exemplary care
Mold JW and Gregory ME Best practices research. Family Medicine 2003, 35 (2): 131-134
Methods (cont.)• Dr. Mold visited each physician and presented
the six principles and• The project provided them with a practice
enhancement assistant (PEA) to assist with implementation• In the practice approximately 1/2 day every
week for 4 months• They were also provided with feedback
from the pre-intervention audits
Methods (cont.)• We also made available a PDA Diabetic
Patient Tracking application conceived of by an OKPRN physician and developed by us prior to this project
• Prompts the nurse (or physician) to follow guidelines
• Creates an auditable registry of diabetic patients
• Produces a flow sheet for the medical record
Principles Derived from Exemplar Interviews Diabetes visits EVERY 3 months for every
diabetic patient Label diabetic charts with sticker Protocol for office staff (triggered by
sticker) Keep a registry of all diabetic patients Work with one or two eye doctors who are
faithful about sending reports and recalling patients
Flow sheet for chart
Results (Process Measures)• High rate of acceptance of six principles
• Mean of 4/6 principles implemented
• High acceptance of the PDA-based diabetic registry
• 21/30 decided to to use it
Results (Outcome Measures)
All diabetic patients > 50 y.o. seen during that 3 month period (pre- or post-intervention) and followed for at least 1 year
• 25 physicians• 595 pre-intervention patients• 582 post-intervention patients
Quality of Care Indicators A1c: 87% 96% p=0.0003 UA protein: 53% 64% p=0.05 Lipid Panel: 69% 80% p=0.02 Foot Exam: 71% 82% p=0.004 Retinal Exam: 48% 59% p=0.04 Pneumovax: 42% 61% p=0.0006 ACEI for BP: 72% 86% p=0.03 ACEI for prot: 53% 64% p=0.05
Paired t-tests; physician as unit of analysis
Comparison Groups Mammography rates unaffected by the
intervention
OFMQ benchmarking study (feedback plus a reasonable performance target based upon 90th percentile of peer performance) showed no significant improvements in DQIP indicators in a similar group of practices the previous year
Conclusions Significant short-term improvement
in physician performance with instruction in principles derived from exemplars plus assistance of a PEA
High level of physician acceptance of the exemplar principles and the PEA
Limitations• Historical control
• Others have reported benefit of benchmarking
• Short term follow-up• Can’t separate individual components of
the intervention • Exemplar principles• PEA• PDA application
Project #4
RCT to Determine Relative Effectiveness of Feedback/Benchmarking, Best Practice Principles, and PEAs
• Three arms with 8 practices in each arm• Audit/feedback/benchmarking (FB)• FB + Best Practice Principles (BPP)• FB + BPP + Practice Enhancement Assistant
(PEA)
Methods FB FB+BPP FB+BPP+PEA
Clinics 8 8 8Clinicians Pre- 14 14 10 Post- 11 14 10Patients Pre- 474 332 387 Post- 481 372 315
Results
A1c in 1 yr (mean A1c)
Pre- Post-FB 71% (7.4) 94% (6.7)FB/BPP 87% (7.9) 85% (7.4)FB/BPP/PEA 75% (7.2) 83% (7.1)
Results
Lipids in 1 yr (mean LDL)
Pre- Post-FB 54% (111) 81% (102)FB/BPP 64% (114) 70% (110)FB/BPP/PEA 66% (104) 71% (106)
Results
Foot exam/1 yr (eye exam/1 yr.)
Pre- Post-FB 59%(35%) 63%(56%) FB/BPP 74% (55%) 61% (59%)FB/BPP/PEA 62% (41%) 39% (44%)
Results
Taking an ACEI Pre- Post-FB 57% 66%FB/BPP 65% 67%FB/BPP/PEA 61% 51%
Results
Pneumovax ever
Pre- Post-FB 20% 46% FB/BPP 56% 54%FB/BPP/PEA 39% 42%
Results
Degree of practice implementation (degree of personal implementation) 1–10 scale
FB 8.2 (8.6)FB/BPP 5.2 (5.9) FB/BPP/PEA 7.4 (7.1)
Results
Difficulty for practice with implementation (personal difficulty) 1-10 scale
FB 5.2 (4.3) FB/BPP 6.5 (5.7) FB/BPP/PEA 4.3 (3.9)
Results
Satisfaction with practice’s management of diabetics (your management) 1-10 scale
Pre- Post-FB 6 (6.4) 8 (8.2)FB/BPP 5 (6.1) 6.5 (7.2) FB/BPP/PEA 5.4 (5.5) 7.9 (8)
Conclusions There was some improvement in
performance overall in all groups Audit/feedback/benchmarking alone
may have worked as well or better than with addition of best practice principles and a PEA
Why????
Speculations
1. Small numbers/randomization failurea) Different levels of motivation/readiness to
changeb) Different levels of ability to change/control
over processes
2. FB Group paid more attention to their audit results and knew they were going to have to address them without help
3. PEAs used ineffective techniques
Questions/Reference