Symposium 1: EBM

Diabetes Care
Prevention

Jonathan Ross, MD
Karen Odato, CNM, MSN, MSLIS
Cindy Stewart, MLS
 The History of Medicine

2000 B.C. - Here, eat this root.

1000 A.D. - That root is heathen. Here, say this prayer.
1850 A.D. - That prayer is superstition. Here, drink this
potion.
1940 A.D. - That potion is snake oil. Here, swallow this pill.
1985 A.D. - That pill is ineffective. Here, take this
antibiotic.
2000 A.D. - That antibiotic doesn't work anymore. Here,
eat this root.

From the Cochrane Collaboration Consumer Network Newsletter

(September, 1999, page 10)
 EBM: What Is It?

“..integrating individual clinical expertise

with the best external clinical evidence
from systematic research.”

“conscientious, explicit and judicious use

of current best evidence in making
decisions about the care of individual
patients.”
Sackett, DL. BMJ. 1996 Jan 13;312(7023):71-2
 What we’re going to cover

• Review selected principles of evidence-

based medicine
– Focus on diabetes care, and
screening and prevention
– Look at the measures and
expressions of risk reduction
 What we’re going to cover

• Review major EBM resources that answer specific

clinical questions:
– Cochrane
– DARE
– ACP Journal Club
• Find the best evidence in MEDLINE.
• Review evidence-based tools that answer general
clinical questions:
– UpToDate
– eMedicine
– National Guideline Clearinghouse
• Review resources that answer drug-related
questions
 Case Presentation

• BG, a 51 yo mother of 3 presents to your office

complaining of polyuria, polydipsia, and weight
gain of 15 lbs over 3 months. Her mother had
DM. The last two of her children were large for
gestational age; the last pregnancy was
notable for pre-eclampsia.
 Questions

• Patient care questions • Teaching improvement

– What is the likelihood questions
that someone with – When a patient presents
gestational diabetes to my office, what kind of
will develop teaching should my
established diabetes? learners (students,
residents) already have
had?

• Practice improvement
questions
– Is my patient’s database
easily retrievable?
 Case, continued

• Her past history is notable for mild diet treated

hypertension. She had a TAH-BSO 5 years ago
for fibroid related menorrhagia. She has a
seizure disorder and takes Dilantin®.
 Question

• Does Dilantin® have any impact on glucose

metabolism?
 Clinical Pharmacology Online
http://cponline.hitchcock.org/

• Clinical Pharmacology is a drug information

application that provides peer reviewed,
clinically-relevant information on drugs
available in the United States, including off-
label uses and dosages, herbal supplements,
nutritional products, and new and
investigational drugs.
 Case, continued

• A quick physical • Initial labs demonstrate a

examination reveals a random blood sugar of
woman appearing her age, 426 mg/dL. There is an
overweight and in no anion gap of 12 and the
evident distress. BUN/Cr are 25 and 1.3
• VS: 155/80 P 96 mg/dL, respectively. The
T 36.5C Height 5’3” urine microalbumin is
Weight 165 lbs No 100 mcg/dl. The
retinopathy/neuropathy hemoglobin A1c is 12.6%
 Question

• Should protein intake be restricted in a middle-

aged patient with Type 2 diabetes and
microalbuminuria?
 The Major EBM Resources to Answer
Specific Clinical Questions

• The Cochrane Database of Systematic Reviews

• The ACP Journal Club
• The Database of Abstracts of Reviews of
Effectiveness (DARE)

• MEDLINE
 The Cochrane Database of
Systematic Reviews - via Ovid
http://www.dartmouth.edu/~biomed

• Published by the International Cochrane

Collaboration. Updated quarterly.

• Consists of detailed, structured topic reviews of

hundreds of articles.

• Teams of experts complete comprehensive

literature reviews, evaluate the literature, and
present summaries of the findings of the best
studies.
 The ACP Journal Club – via Ovid
http://www.dartmouth.edu/~biomed

• Electronic access to articles in The ACP Journal

Club, published bimonthly by the ACP-ASIM.

• The editors of this journal screen the top 100+

clinical journals and identify studies that are
methodologically sound and clinically relevant.

• An enhanced abstract, with conclusions clearly

stated, and a commentary are provided for
each selected article.
 Database of Abstracts of Reviews of
Effectiveness (DARE) – via Ovid
http://www.dartmouth.edu/~biomed

• Produced by the National Health Services'

Centre for Reviews and Dissemination (NHS
CRD) at the University of York, England.

• Contains structured abstracts of systematic

reviews from a variety of medical journals.

• Updated monthly.
 EBM Reviews

• Ovid allows you to search Cochrane, ACP

Journal Club, and DARE simultaneously.

• From the Ovid database list, select “EBM

Reviews – full-text”
 Hypothetic Examples of RR, ARR, NNT
Measures in 4 Studies
Group # Pts # Events RR ARR NNT
Placebo 1000 1 50% 0.05% 2000
Treated 1000 .5

Placebo 1000 10 50% 0.5% 200

Treated 1000 5

Placebo 1000 100 50% 5% 20

Treated 1000 50

Placebo 1000 1000 50% 50% 2

Treated 1000 500
 Limitations of NNT

• NNT indicates frequency, not utility

• NNT is based on an outcome for a specified
period, with treatment delivered in a specified
way
• NNT should not be compared across conditions
• NNT assumes that a given intervention
produces the same relative risk reduction
exclusive of baseline risk
• Each NNT has a confidence interval
 p values or confidence intervals?

• p values test the evidence against a null

hypothesis (e.g., p=0.05 means we can be
sure the hypothesis tested is likely to be true
95% of the time.)

• Confidence intervals tell us about the strength

of the evidence (e.g., 95% CI is the range of
values within which we are 95% sure that the
true value lies.)
 MEDLINE via Ovid*
http://www.dartmouth.edu/~biomed
• The National Library of Medicine’s premier database
covering the fields of medicine, nursing, dentistry,
veterinary medicine, the health care system, and the
preclinical sciences.

• Contains bibliographic citations and author abstracts

from more than 4,600 biomedical journals published in
the United States and 70 other countries.

• Over 11 million citations dating back to the mid 60s.

Updated weekly on the Ovid system.

*Requires Kerberos authentication for access

 Another Specific Question

• How does the A1c correlate with average

blood sugar?
 Defining the Relationship Between Plasma Glucose and HbA1c: Analysis of glucose profiles and
HbA1c in the DCCT
Diabetes Care 2002;25(2):275-278

A1c vs plasma glucose

• Erythrocyte life span 120 days
350 • Recent (3-4 weeks) PG levels
contribute more (50%) than
300 remote (90-120 d) to A1c
250 • FPG tends to underestimate
A1c; post-lunch PG correlates
200
MPG well with MPG.
150 (mg/dL) • 1% change in A1c correlates
100 with  MPG 35 mg/dL

50
0
5 6 7 8 9 10 11 12
A1c (%)
 Question

• The DCCT was a trial in type 1 diabetes. But I

remember there was a UK study on A1C in
type 2 diabetes; how can I find it fast?
 MEDLINE via PubMed

• Free access to MEDLINE from the National

Library of Medicine

• Includes links to Dartmouth’s digital full-text

journals, when available
– [Note: must connect to PubMed through the
Biomedical Libraries Web to access full-text]
 Intensive blood-glucose control with sulphonylureas or
insulin compared with conventional treatment and risk of
complications in patients with type 2 diabetes (UKPDS 33).
Lancet. 1998
Sep 12;352(9131):837-53.
• RCT, median f/u 10 years
50
• 23 hospital based clinics in UK
45
Clinical • 3867 pts, mean age 53, 61%
40 events/1000pt-y men, BMI 27.5, newly diagnosed
35 after 3 mo diet therapy
30 A1c- % • 2729 pts intensive therapy
25 • 1138 pts conventional dietary
20 Hypoglycemia • A1c 7.0% v 7.9%
15 %/year • Hypoglycemia 0.7% v 1-1.8%
• Mortality- NS
10
• Macrovascular- NS
5
0
Intensive

1% reduction in A1c subsequently

associated with 35% reduction in microvascular endpoints,
18% reduction in MI, 17% reduction in all cause mortality
 Randomized Controlled Trial

• The ideal population based strategy (actually, the ideal

agricultural strategy)
• Tests the average efficacy of (therapeutic) interventions
• Relies on double blind methodology rather than
sophisticated knowledge of clinical variables
• Biases: TNTC
 The Randomized Controlled Trial

Disease
Patients Present Absent

Treatment A B EER= A/A+B

Control C D CER= C/C+D

EER= experimental event rate ARR (absolute risk reduction) = CER-EER

CER= control event rate RRR (relative risk reduction) = ARR/CER
NNT (number needed to treat) = 1/ARR
 Case, continued

• This brief office visit has established that Ms.

BG has new onset diabetes. Initial plans are
formulated, including a visit to a nurtritionist.
A decision needs to be made concerning
additional treatment. She has significant
problems concerning finances. After
considering the patient’s clinical status, her
resources and ability to return for care, an Rx
for metformin is considered.
 Question

• What’s current information on the

management of DM Type 2 with oral
glycemics? If I do initiate treatment with an
oral agent, which one should it be?
 Evidence-based Tools to Answer
General Clinical Questions

• Textbooks
• UpToDate
• eMedicine, et. al.
• National Guideline Clearinghouse

• MEDLINE
 UpToDate
http://www.dartmouth.edu/~biomed
http://uptodate.com

• Topic reviews are written by recognized

authorities who review the topic, synthesize
the evidence, summarize key findings, and
provide specific recommendations.

• Physician editors and authors review and

update the content on a continuous basis; a
new, peer-reviewed version is issued every
four months.
 eMedicine
http://www.emedicine.com

• Nearly 10,000 physician authors and editors

contribute to the eMedicine Clinical Knowledge
Base with coverage of 7,000 diseases and
disorders.
• All of eMedicine's original content undergoes
four levels of physician peer review plus an
additional review by a PharmD.
• Contains an Image Bank of nearly 30,000
multimedia files.
• Updated daily.
 Another General Question

• Is there a guideline describing the

management of someone with type 2 diabetes
(e.g., frequency of visits, labs, etc.)?
 The National Guideline Clearinghouse
http://www.guideline.gov

• A comprehensive database of more than 995

evidence-based clinical practice guidelines and
related documents.

• Sponsored by the Agency for HealthCare

Research and Policy in partnership with
– The American Medical Association
– The American Association of Health Plans

• Updated weekly.
 Back to some Previous Questions

• How can I access drug cost information

quickly?

• Before I prescribe Metformin: are there any

known interactions between it and Dilantin?
 Handheld PDA Resources

http://www.dartmouth.edu/~biomed/services.ht
mld/pda.resources.shtml
 Case, continued

• Ms. BG had many • She had watched her

questions. She mother getting pain
wanted to know what in her feet, and also
the dangers of developing problems
having diabetes with her vision that
were, and what she required laser
could do to reduce treatment.
her chances of
getting them.
 Question

Are there patient education materials that would

help her better understand and manage her
disease?
 MEDLINEplus
http://medlineplus.gov/

• Extensive information from the National Institutes of

Health and other trusted sources on over 600 diseases
and conditions.

• Also includes lists of hospitals and physicians, a medical

encyclopedia and a medical dictionary, health
information in Spanish, extensive information on
prescription and nonprescription drugs, health
information from the media, and links to thousands of
clinical trials.

• Updated daily.
 Other consumer health resources

• http://www.dartmouth.edu/~biomed/resources
.htmld/conshealth.htmld/
 Case, continued

• Ms. BG returned 2 weeks later. She felt better

and was no longer having polyuria. During the
visit she stated that this was a huge wake-up
call to her, and she wanted to take excellent
care of herself. The nutritionist and she had
worked out a good plan of diet and exercise.
She wanted to address heart disease risk and
cancer prevention as well.
 Prevention Topics

– Breast cancer
– Colon cancer
– Prostate cancer
– Cervical cancer
– Cardiovascular
• Cholesterol
• Homocysteine
• CRP
 Efficacy of Mammography-
Women Under 50
Study RR ARR NNS
HIP .778 .00062 1,606
Malmo 1.326 -.00005 -1,938
S2C 1.131 -.00013 -7,803
Edinburgh .987 .00003 34,248
Stockholm 1.025 -.00003 -36,143
Canada 1.35 -.0004 -2,521

Total 1.02 -.000034 -29,565

Based on 119/79,103 deaths in control- 140/91,016 deaths in

screened, 29,565 women would need to be screened to
‘cause’ one death.
 Efficacy of Mammography-
Women Over 50

Study RR AR NNS

HIP .604 .00155 645

S2C .613 .00087 1,151
Malmo .680 .00062 1,619
Edinburgh .810 .00075 1,335
Stockholm .530 .00082 1,217
Canada .974 .000052 19,069
Total .655 .00089 1,122

Baseline risk of death from breast cancer in this age group is

0.00271.
Based in 301/116,387 deaths in control group - 247/145,711 deaths
in screened group = AR

NNS to prevent one death from breast cancer is 1122.

Breast cancer screening with mammography
1000 women

8 with breast cancer 992 without breast cancer

7 test positive 1 test negative 70 test positive 922 test negative

Thus the probability of having cancer when the test is positive

is 7/77, or 9.1%
 WHI JAMA;2002;288:321-33
16,608 women aged 50-79 (mean 63.3) RCT 5.2 yrs
Event rates per patient-year

Outcomes HRT Placebo RRI(R) NNH (T)

CHD 0.37% 0.30% 29% (2-63) 1152 (531-16693)

VTE 0.34 0.16 111% (58-182) 565 (345-1079)

Breast Ca 0.38 0.30 26% (0-59) 1285 (567-infinity)

Hip Fx 0.10 0.15 34% (2-55) 1962 (1213 -33358)

Colorectal 0.10 0.16 37% (8-57) 1691 (1097 -7819)

Ca
 HRT use in 10,000 Women
Benefits and Harms/ Year
(from JAMA 2002;288:872-881)

EventsHRTPlaceboExcess (Fewer)CHD37307VTE341618Breast Ca3830

 Q: Does FOBT reduce the incidence of CRC?
NEJM. 2000;343:1603-7

• 46,551 pts, 18 y f/u; 52% female;

Incidence of CRC 91% f/u. 75% compliance
• Incidence of cancer after 18 years
at 18 years
• CER 0.39%, EER 0.32%
• NNT 1428
1.00% • ARR 0.07%
0.80% • RRR 18%
0.60%
0.40%
0.20%
0.00%
Placebo
The Hemoccult problem
10,000
people

30 colorectal cancer 9,970 no colorectal cancer

15 positive 15 negative 300 9,670

positive negative

Thus the probability of having cancer when the test is positive

is 15/315, or 4.8%
 Do statins lower the risk of cardiac events (primary prevention?)
AFCAPS/TexCAPS JAMA, 1998,279:1615-22

• 6605 pts; 85% male; chol 180-264

Composite endpoint mg/dl, HDL <45 mg/dl; f/u 5.2 yrs
• fatal/nonfatal MI, ACS, sudden death
• CER 6%, EER 4%
• NNT 50 (33-97)
6.00%
• RRR 37% (21-50)
5.00%
4.00%
3.00%
2.00%
1.00%
0.00%
Placebo
 In the works….

• The General Internal Medicine Evidence Based

Resource
– Web access
– PDA compatible
 EBM Resources: Summary

• When you’re looking for evidence on which to

base specific patient care decisions, you can read
and fully evaluate every article on your subject of
interest.

• Or you could employ the resources that do the

study reviews and filtering for you:
– The Cochrane Database of Systematic Reviews
– The ACP Journal Club
– DARE
 Summary, contd.

• When reviews are not available, you can craft

search strategies in MEDLINE to limit your
search to the evidence-producing studies.

• For overviews of broader topics, use the tools

that summarize the evidence:
– UpToDate
– eMedicine
– National Guideline Clearinghouse
– et. al.
 Self assessment

• What questions do I • Ask!

want to ask?
• Do I know how to • Access!
access the
resources?
• Assess!
• Do I know how to
interpret the
information? • Apply!
• Can I explain the
answer to a colleague
or patient?
 The Encounter Paradigm

Knowledge Preparedness

Patient

Current
Fidelity
Information You
Accessible Knowledge
Accurate Energy
 Thank you!

jonathan.ross@hitchcock.org
karen.odato@dartmouth.edu
cindy.stewart@dartmouth.edu