Presents: Expert Training Series on Glycemic Control

1
PresentsExpert Training Series on Glycemic
Control

• Jeffrey L. Schnipper, MD, MPH
• Director of Clinical Research
• BWH Academic Hospitalist Service
• Brigham and Womens Hospital
• Assistant Professor of Medicine
• Harvard Medical School

2
Goals for the Evening

• Understand principles of QI
• Review management of inpatient glycemia
• Analyze processes and gaps at your own
  institution
• Learn how to assemble a QI team
• Create general and specific aims for improvement
• Understand measurement and metrics
• Begin development of QI tools, including
  protocols, order sets, and high reliability
  interventions
• Design PDSA cycle 1
• Think about barriers to change

3

• I. Quality Improvement Overview

4
The Evolving Culture of Medicine

• 20th Century Characteristics
• Autonomy
• Solo Practice
• Continuous learning
• Infallibility
• Individual Knowledge

• 21st Century Characteristics
• Teamwork systems
• Group practice
• Continuous improvement
• Multidisciplinary problem solving
• Change

Shine, KI. Acad.Med. 20027791-99
5
Definition of Quality

• Meeting the needs and exceeding the expectations
  of those we serve
• Delivering all and only the care that the patient
  and family needs

6
Traditional Quality Assurance
outliers
7

• Quality Improvement

Before
After
better
worse
Quality
better
worse
Quality
8

• Quality Improvement is
• Focus on processes of care
• Reduced variation by shifting entire practice
• A change in the design of care
• Quality Improvement is NOT
• Forcing people to work harder / faster / safer
• Traditional QA or peer review
• Creating order sets or protocols without
  monitoring use or effect

9
System Based Care

• In highly reliable systems, the system supports
  quality
• E.g computer cords, you cant plug the wrong cord
  into the wrong spot, it wont fit- you can with
  arterial vs. venous central lines
• Every system is designed to get exactly the
  outcome it gets.

10
(No Transcript)
11
(No Transcript)
12
The Multidisciplinary Team the Engine of
Quality Improvement
13
Features of a Good Team

• Safe
• no ad hominem attacks
• Inclusive
• open to all potential contributors
• values diverse views not a clique
• Open
• considers all ideas fairly
• Consensus seeking
• finds a solution all members can support

14
Quality Team Members

• Three kinds of members
• A Team Leader
• A Team Facilitator
• 3. Team Members process owners
• (people with operational, hands-on fundamental
  knowledge of the process)

15
Team Ground Rules

• Members
• Attend regularly
• Are equal
• Speak freely in turn
• Attack problems, not people
• Meetings
• Start and end on time

16
Team Ground Rules

• Consensus
• Honesty before cohesiveness
• All agreements kept unless renegotiated
• We speak with One Voice
• Especially after leaving the meeting
• Silence equals agreement

17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
Plan- Workflow Analysis

• Analyze your system to find how the work is done
• Will likely need input from others involved in
  the process

21
Do-Choose an Intervention

• Find a weak spot in your current processes (a big
  hole in the cheese)
• Identify one that is system based (not personal
  responsibility)
• Try to find the simplest one with the biggest
  impact

22
Study- Types of Measurement/metrics

• System Measure measures capacity (e.g. how many
  computers on a unit)
• Process Measure measures how well you are doing
  something (e.g. rate of influenza vaccines given
  before discharge)
• Outcome Measure - ultimate measure (e.g. rate of
  in hospital mortality)

23
Act- Sustainability

• Project should be ongoing
• Make the intervention(s) part of routine care
• Use data to refine your intervention
• Move to another area or
• Choose another defect in the process

24
A Model for Improvement

• Aims
• Time-specific and measurable
• Defined population

• Measures
• Quantitative

• Changes
• Select interventions most likely to make
  improvements

• Testing Changes
• PDSA cycle scientific method for
  action-oriented learning
• Plan change, try it, observe results act on
  what is learned

25
Features of Good Aim Statements

• Specific
• Measurable
• Aggressive yet Achievable
• Relevant
• Time-bound

26
Sample Aim Statements

• VTE Prevention
• Within the next 6 months 95 of patients will be
  on a VTE prophylaxis regimen appropriate for
  their level of VTE risk, as defined by our
  protocol.
• Within 12 months we will halve the number of
  hospital acquired VTE in our institution.

27
A Model for Improvement

• Aims
• Time-specific and measurable
• Defined population

• Measures
• Quantitative

• Changes
• Select interventions most likely to make
  improvements

• Testing Changes
• PDSA cycle scientific method for
  action-oriented learning
• Plan change, try it, observe results act on
  what is learned

28
MeasuresWhy Are You Collecting Data?

• Understand variation in process
• Monitor process over time
• See effect of change in process
• Provide a common reference point
• Provide accurate basis for predicting future
  performance

29
Where is Quality Measured?
Structure
Process
Outcomes of Care
Inputs
Steps
Outputs

• Patients
• Providers
• Equipment
• Supplies
• Training
• Environment
• Workload

• Physician orders
• Nursing care
• Coordination
• Inventory methods
• Housekeeping
• Transport

• Physiologic
• parameters
• Functional status
• Satisfaction
• Cost
• Mortality
• LOS
• Morbidities

Donabedian, 1988
30
Measurement Principles

• Seek usefulness, not perfection
• Integrate measurement into daily routine
• Use qualitative and quantitative data
• Use sampling
• Plot data over time
• Use a balanced set of measures for all
  improvement efforts

31
Balancing Measures

• Are changes designed to improve one part of
  the system causing new problems in other parts of
  the system?
• Intervention to reduce time spent on ventilator
  after surgery
• Assure reintubation rates are not ?
• Intervention to reduce hospital length of stay
• Assure readmission and mortality rates are not ?

32
II. Quality Improvement Toolkit
33
QI Toolkit

• Methods to complete QI project steps
• Problem identification
• The GAP difference between actual performance
  potential performance
• Data analysis
• Solution planning
• Result evaluation

34
QI Toolkit

• General
• Brainstorming
• Problem Identification techniques
• Process mapping
• Failure modes and effects analysis
• Cause and effect diagrams

• Data analysis / Solution planning
• Affinity diagrams
• Pareto charts
• 2 x 2 tables
• Display metrics
• Run charts
• Statistical process control charts

35
Process Mapping/Flow Charting

• Picture of process function in organization
• Actual process vs. ideal process
• Identifies all work steps
• Show participants, inputs, outputs
• Understand hand-offs track resources
• Shapes used as symbols
• Diamond yes/no decision
• Oval input/outputs
• Box task performed
• Arrow direction

36
Workflow Analysis
37
Failure Mode and Effects Analysis (FMEA)

• Everything that can fail, shall fail
• 
  Murphys law
• Prospective risk analysis
• Fix it before it breaks
• Ask
• What could go wrong?
• How badly could it go wrong?
• What can be done to prevent the failure?
• Prioritize actions according to quality cost

38
FMEA - Steps

1. Identify process / technology involved
2. Form multidisciplinary team
3. Identify failure modes
4. Identify causes of failure modes
5. Identify effects of failure modes
6. Calculate criticality
7. Recommend actions
8. Monitor results

39
FMEA Failure Mode Identification

• Identify potential or actual errors for each
  process step
• Use internal error reporting data and external
  reports
• Ask, What if?
• Identify the causes of failure modes
• Identify the effects on the patient and staff
  involved

40
Failure Mode Analysis
Long delay in triage
Fail outpatient appointment
Long delay to cath lab
41
Cause Effect Diagram Process

• Identify define the problem (effect)
• Place the problem at head of diagram
• Identify the broad causal areas of the problem
• Systems view humans, technology, policy
  procedure, resources, environment
• List the possible causes in each broad area
• Evaluate each cause for more specific causes
• Select main causes from diagram
• Set goals take action on main causes

42
(No Transcript)
43
2 x 2 Tables

• Prioritize data or solutions to identify
• Biggest bang for buck
• Maximize potential for change resources used
• Low-hanging fruit
• Easy interventions (regardless of impact)
• Develop 2x2 table
• Yield (high/low) x feasibility (high/low)
• Can use other priorities (cost, urgency, etc)
• Place ideas into categories for prioritization

44
2 x 2 Table
Feasibility
High
Low
High
DO NOW
PLAN TO DO
Yield
Low
RESIST CEASE
DO NOW / NEXT
45
Creating a Run Chart

• Organize data
• Divide data into two sets of values X Y
• X values represent time
• Y values represent the measurements taken
• Chart data
• Plot the Y values versus the X values
• Plot the median value for the data
• Add benchmark / comparison lines

46
Understanding Variation

• Variation exists in everything we do
• Systems processes are not static
• Lack of understanding leads to
• Seeing trends where there are no trends
• Make wrong conclusions about variation source
• Blame or credit given to staff for processes not
  in their control
• Cannot predict future or make improvements in
  processes

47
Statistical Process Control Charts

• Run chart with statistically determined upper and
  lower control limit lines
• Limits are usually 3 standard deviations from the
  data mean
• Standard deviation measure of data variability
• 1 SD 68 data 2 SD 95 3 SD 99

48
Stable Process but Considerable Variance
Common cause variation
Special cause variation
49
Steady Improvement and Reduced Variation
50
Steady Improvement Intervention is Working
Good special cause variation
51
III. Quality Improvement Standardization and
Reliability
52
Definitions

• Reliability is defined as failure free operation
  over time.
• Reliability is measured as the inverse of the
  systems failure rate

Improving the Reliability of Healthcare, IHI
Executive Summary 2004
53
Health Care Reliability Where is Current State?

• 10-1 1 or 2 failures out of 10 opportunities
• 10-2 5 failures or less out of 100
  opportunities.
• 10-3 5 failures or less out of 1000
  opportunities.
• 10-4 5 failures or less out of 10,000
  opportunities.
• 10-6 Aviation and Nuclear Power Plants
• defect rate one in 1,000,000
• Chaotic Process Failure in gt 20 of the time

54
Failure to Do Simple Things Well

• Wash Hands
• 60 Reliable
• Patients Understand Meds / Problems
• 30 Reliable
• Central Lines Placed w/ Proper Technique
• 60 Reliable
• Basal Insulin for Input Uncontrolled DM
• 40 Reliable
• VTE Prophylaxis
• 30-50 Reliable

55
Inherent Human Limitations

• Limited memory capacity 5-7 pieces of
  information in short term memory
• Negative effects of stress error rates
• Tunnel vision
• Negative influence of fatigue and other
  physiological factors
• Limited ability to multitask cell phones and
  driving

56
How Reliably Do Defects Occur?

Salvendy and Park Compendium of Human Factors
57
Failure Rates Are Predictable
Probability of success, each element
0.99
0.999
0.9999
0.95
of elements
170 interventions per day in the ICU
58
Making Our Care Environment Reliable

• We cant change the human condition but we can
  change the conditions under which humans work.
• Reason J. BMJ 320(7237)768-770

59
Approaches to Reliability

• 10-1 strategy
• prevent failure by following uniform processes
• intent, vigilance, hard work model
• 10-2 strategy
• identify and mitigate failures
• use of human factors engineering
• 10-3 strategy
• identify and correct problems in real time

60
The VTE Journey at 10-1

• Typical Standardization
• Experts meet to design protocol
• Protocol presented as final product
• Practitioners vary from protocol for appropriate
  and inappropriate reasons
• Variation only noticed when bad outcome
• Protocol does not improve, no collective wisdom
• Protocol exists but not integrated into
  workflow/order sets
• Reliability 60-85

61
10-2 Tier

• Learn from variation, oversights, failures
• Variation from protocol allowed but monitored and
  examined
• Variation from protocol based on different
  patient situations incorporated into the protocol
• variations without justification, based on
  physician variability, become apparent
• Defects are used to move to a learning
  system---tweak, tweak, tweak

62
10-2 Strategies Integrating Human Factors

• Desired action is
• the default (MD must opt out)
• Prompted by a reminder or decision aid, i.e.
  algorithms incorporated into orders
• Standardized into a process, i.e. admit orders
• Scheduled to occur, i.e. pharmacist review of
  orders for DVT prophylaxis every 3 days
• Responsibilities for desired action are redundant

63
10-2 Strategy Decision Support Tools

• Incorporate Algorithm into Standardized Order
  Sets
• Provide decision tree for diagnosis and/or
  therapy to guide choices
• Risk assessment
• Diagnosis
• Therapy
• Reduces variability and works best for majority
  of cases
• Education needed
• When should I intelligently vary from the
  algorithm based on this patients unique
  circumstances?

64
10-3 and Beyond. . .

• Involves detection and mitigation strategies in
  real time
• Rate of fatal adverse events for anesthesiology
  and blood transfusion reduced to 10-5
• Detection functions such as double checks, pulse
  ox, barcoding
• Automation and forcing functions
• Barcoding, CPOE

65
Key Principles of Effective Interventions /
Protocols

• Pilot on a small scale
• Be mindful of the workflow
• Monitor use of the protocol
• Allow for variation from protocol based on
  patients (not providers) and improve protocol
  based on feedback and justifiable variation
• Implementation of a reliable intervention is an
  ongoing process

66
Hierarchy of Reliability
Predicted Reliability
Level
40
1

• No protocol (State of Nature)
• Decision support exists but not linked to order
  writing, or prompts within orders but no decision
  support
• Protocol well-integrated (into orders at
  point-of-care)
• Protocol enhanced (by other QI and high
  reliability strategies)
• Oversights identified and addressed in real time

50
2
3
65-85
4
90
5
95
67
The Name of the Game

• Think like an engineer and design more
  Reliability into processes
• Make it easy (and automatic) to do the right
  thing!

Perceived Burden
Noncompliance
Perceived Risk
Clapper, C. ICSI/IHI Colloquium, May 16, 2007
68
Special Thanks to

• Tosha Wetterneck
• Lakshmi Halasyamani
• Erin Stuckey
• Janet Nagamine

69

• Review
• Using Subcutaneous Insulin to Reliably Achieve
  Desired Glycemic Outcomes in Non-Critically Ill
  Hospitalized Patients

70
Objectives For This Lecture

• Appreciate the obstacles to achieving good
  glycemic control in hospital patients
• Understand and apply the best practice of
  inpatient hyperglycemia/diabetes management using
  subcutaneous insulin
• Understand the common deviations from the best
  practices of insulin management in the hospital
• Learn how to prevent and manage hyperglycemia and
  hypoglycemia

71
Case 1

• 56 year old woman with DM2 admitted with a
  diabetes-related foot infection which may require
  surgical debridement in the near future, eating
  regular meals.
• Weight 100 kg
• Home medical regimen Glipizide 10 mg po qd,
  Metformin 1000 mg po bid, and 20 units of NPH q
  HS
• Control A recent HbA1c is 10, POC glucose in ED
  240 mg/dL
• What are your initial orders for basal and
  nutritional insulin?
• How would you manage the oral agents?

72
Managing Diabetes in the Hospital Presents
Different Challenges than Managing Diabetes in
the Outpatient Arena!

• The hospital is associated with
• Nutritional and clinical instability
• The need for changes from the home diabetes
  medical regimen
• Acute illness, stress-related hyperglycemia
• Use of medications that impact glycemic control

73
Why Should We Care?

• Hyperglycemia occurs frequently in hospital
  patients, and is associated with poor outcomes
• Hypoglycemia occurs frequently in hospital
  patients, and is unpleasant and dangerous
• Adequate metabolic control is an attainable goal
  for hospital patients

74
What is the Appropriate Glycemic Control Target
for Inpatients?

• Controversial!

ICU Non-ICU, Preprandial Non-ICU, Maximum
ACCE/ACE 110 mg/dl 110 mg/dl 180 mg/dl
ADA 110 mg/dl 90-130 mg/dl 180 mg/dl
75
How Can Diabetes and Hyperglycemia be Controlled
in the Hospital?

• Oral agents often inappropriate for hospital
  patients
• IV insulin most often used in the intensive
  care unit setting (or in other defined
  populations)
• Subcutaneous insulin the drug of choice for
  controlling hyperglycemia in the majority of
  non-critically ill patients

76
Oral Antidiabetes Agents in the Hospital

• Oral agents can be continued in stable patients
  with normal nutritional intake, normal blood
  glucose levels, and stable renal and cardiac
  function. However, there are several potential
  disadvantages to using these medications in
  hospital patients
• Disadvantages of most oral agents
• Slow-acting/difficult to titrate
• Disadvantages of insulin secretagogues (e.g.
  sulfonylureas and meglitinides such as glyburide,
  glypizide, repaglinide, etc.)
• Hypoglycemia if caloric intake is reduced
• Some are long-acting (hypoglycemia may be
  prolonged)
• Disadvantages of metformin
• Lactic acidosis can occur when used in the
  setting of renal dysfunction, circulatory
  compromise, or hypoxemia
• Slow onset of action
• GI complications Nausea, diarrhea
• Disadvantages of thiazoladinediones (e.g.
  rosiglitazone, pioglitazone)
• Slow onset of action (2-3 weeks)
• Can cause fluid retention (particularly when used
  with insulin), and increase risk for CHF

77
Indications for IV Insulin Therapy

• Prolonged fasting (gt12 h) in type 1 DM
• Critical illness
• Before major surgical procedures
• After organ transplantation

• DKA
• Labor and delivery
• Acute MI
• Other illnesses requiring prompt glucose control

ACE Position statement on inpatient diabetes 2004
78
What is the Best Practice for Managing Diabetes
and Hyperglycemia in the Hospital?

• The answer is anticipatory, physiologic insulin
  dosing, prescribed as a basal/bolus insulin
  regimen
• This means giving the right type of insulin, in
  the right amount, at the right time, to meet the
  insulin needs of the patient
• Not Sliding Scale Insulin

79
The Components of a Physiologic Insulin Regimen

• Basal insulin
• Nutritional insulin
• Correctional insulin

80
Physiologic Insulin Secretion Basal/Bolus
Concept
Nutritional (Prandial) Insulin
50
Insulin (µU/mL)
25
0
Basal Insulin
Breakfast Lunch Supper
150
Nutritional Glucose
The 50/50 Rule
100
Glucose (mg/dL)
50
Basal Glucose
0
7
8
9
10
11
12
1
2
3
4
5
6
7
8
9
A.M.
P.M.
Time of Day
81
The Components of a Physiologic Insulin Regimen

• Basal insulin - long-acting insulin required in
  all Type 1 (and most Type 2) patients to maintain
  euglycemia by preventing gluconeogenesis
• Nutritional insulin - scheduled short-acting
  insulin given just before a meal, in anticipation
  of the glycemic spike that occurs due to
  carbohydrate ingestion (this dose is given even
  when the blood sugar is in the normal range).
• Correctional insulin - short-acting insulin that
  is given in addition to scheduled nutritional
  insulin (or given at other times of the day) as a
  response to preexisting high blood glucose
  levels

82
Providing Exogenous Basal Insulin

• Long-acting, non-peaking insulin is preferred as
  it provides continuous insulin action, even when
  the patient is fasting
• Required in ALL patients with type 1 diabetes
• Many patients with type 2 diabetes will require
  basal insulin in the hospital
• Can be estimated to be about 1/2 of the total
  daily dose of insulin (TDD)

83
Which Insulins are Best for Basal Coverage?
Glargine (Lantus)
Regular
Lispro (Humalog) Aspart (Novolog) Glulisine
(Apidra)
Insulin Effect
Inhaled insulin
0
6
12
18
24
Time (hours)

• Glargine or detemir are preferred
• NPH also possible

84
Providing Exogenous Nutritional Insulin

• Usually given as rapid-acting analogue (preferred
  in most cases) or regular insulin, for those
  patients who are eating meals
• Must be matched to the patients nutrition
• Should not be given to patients who are not
  receiving nutrition (e.g. NPO)
• Can be estimated to be about ½ of the total daily
  dose of insulin (TDD)

85
Which Insulins are Best for Basal Coverage?
Glargine (Lantus)
Regular
Lispro (Humalog) Aspart (Novolog) Glulisine
(Apidra)
Insulin Effect
Inhaled insulin
0
6
12
18
24

• Rapid-acting insulin is preferred when patients
  are eating meals
• Regular insulin also possible, esp. for tube feeds

Time (hours)
86
Providing Exogenous Correctional Insulin

• Correctional insulin is extra insulin that is
  given to correct hyperglycemia
• Usually rapid-acting or regular insulin (usually
  the same as the nutritional insulin)
• Often written in a stepped format that is used
  in addition to basal and nutritional insulin
• Customized to the patient using an estimate of
  the patients insulin sensitivity
• If correctional insulin is required consistently,
  or in high doses, it suggests a need to modify
  the basal and/or nutritional insulin doses

87
Which Patients Should be Treated with a
Physiologic Insulin Regimen?

• During hospitalization
• Any patient with blood glucose levels
  consistently above the target range
• Immediately at the time of admission
• All patients with type 1 diabetes
• Patients with type 2 diabetes if
• They are known to be insulin-requiring
• They are known to be poorly controlled despite
  treatment with significant doses of oral agents
• They are known to require high doses of oral
  agents that will be held in the hospital

88
A Stepwise Approach to Physiologic Insulin Dosing
in the Hospital

• Decide if patient is appropriate for the protocol
  and discontinue oral antidiabetic agents
• Calculate the estimated total daily dose (TDD) of
  insulin
• Determine the distribution of the TDD calculated
  above based on nutrition regimen.
• Re-evaluate adjust the TDD daily based on the
  glycemic control of the previous 24h

89
Estimate the Amount of Insulin the Patient Would
Need Over One Day, If Getting Adequate Nutrition
Total Daily Dose (TDD)

• Insulin drip-based estimate
• For patients already treated with insulin,
  consider the patients preadmission subcutaneous
  regimen and glycemic control on that regimen
• Weight-based estimate
• TDD 0.4 units x Wt in Kg
• Adjust down to 0.3 units x Wt in Kg for those
  with hypoglycemia risk factors, including kidney
  failure, type 1 diabetes (especially if lean),
  frail/low body weight/ malnourished elderly, or
  insulin naïve patients
• Adjust up to 0.5-0.6 units (or more) x Wt in Kg
  for those with hyperglycemia risk factors,
  including obesity and high-dose glucocorticoid
  treatment

90
Decide Which Components of Insulin the Patient
Will Require, and Which Percentage of the TDD
Each Should Represent

• Basal insulin can generally be estimated to be
  1/2 of the TDD
• Nutritional insulin makes up the remaining 1/2 of
  the TDD

91
Assess the Patients Nutritional Situation

• Eating meals or receiving bolus tube feeds
• Eating meals but with unpredictable intake
• Getting continuous tube feeds
• Getting tube feeds for only part of the day
• Getting parenteral nutrition
• NPO

92
Assess Blood Glucoses at Least Daily, Adjusting
Insulin Doses as Appropriate

• Blood glucose targets can only be achieved via
  continuous management of the insulin program
• There is no autopilot insulin regimen for a
  hospitalized patient!

93
Assess Blood Glucoses at Least Daily, Adjusting
Insulin Doses as Appropriate

• There is no consensus on the best way to titrate
  insulin
• Add up previous days TDD
• Include amount given as supplemental insulin?
• Adjust up or down depending on degree of hypo-
  and hyperglycemia
• Adjust for other factors (renal function,
  steroid dose, nutritional intake, severity of
  illness)?
• Divide new TDD into basal and nutritional
  components
• Split 50/50?
• Adjust basal and nutritional separately,
  depending on AM fasting vs. mid-day sugars?
• Split, but keep AM vs. mid-day sugars in mind?

94
Back to Case 1

• 56 year old woman with DM2 admitted with a
  diabetes-related foot infection which may require
  surgical debridement in the near future, eating
  regular meals.
• Weight 100 kg
• Home medical regimen Glipizide 10 mg po qd,
  Metformin 1000 mg po bid, and 20 units of NPH q
  HS
• Control A recent HbA1c is 10, POC glucose in ED
  240 mg/dL
• What are your initial orders for basal and
  nutritional insulin?
• How would you manage the oral agents?

95
Case 1 Solution

• Bedside glucose testing AC and HS
• Discontinue oral agents
• Total daily dose 100 kg x 0.6 units/kg/day 60
• Basal Glargine 30 units q HS
• Nutritional Rapid-acting analogue 10 units q ac
  at the first bite of each meal
• Correction Rapid-acting analogue per scale q ac
  and HS (Note Use correctional insulin with
  caution at HS, reduce the daytime correction by
  up to 50 to avoid nocturnal hypoglycemia)

96
(No Transcript)
97
Key Review Articles

• Inzucchi. Management of Hyperglycemia in the
  Hospital Setting. N Engl J Med 20063551903-11.
• Clement and colleagues. Diabetes Care 2004 27
  553-91.
• American College of Endocrinology Position
  Statement on Inpatient Diabetes and Metabolic
  Control. Endocrine Practice 2004 10 77-82.
• American College of Endocrinology and American
  Diabetes Association Consensus Statement on
  Inpatient Diabetes and Glycemic Control. Diabetes
  Care 2006 29 1955-62.

98
Special Thanks To

• Greg Maynard, MD
• Kendall Rogers, MD