Using Computers for EvidenceBased Medicine: Diagnosis

1
Using Computers for Evidence-Based Medicine
Diagnosis Therapy

• Terry Shaneyfelt , MD, MPHDivision of General
  Internal Medicine
• EBM Noon Conference Series-I
• July 17, 2007

2
Objectives

• Determine when which resources to use to answer
  different types of clinical questions
• Become familiar with evidence resources available
  at Lister Hill Library
• Understand the role of evidence in making
  clinical decisions

3
Disclaimer

• Lots of resources available
• I havent used them all extensively
• Biased
• Like what I use
• Use what I like
• Suggestions from my EBM geek friends
• Not going to review packaged resources
• E.g. EBSCO host, MD Consult, etc

4
Background

• Point-of-care EBM tools proliferating
• Deciding on right tool(s) is difficult as
  products vary in
• Complexity
• Content
• Accessibility
• Intended audience

5
Background-2

• What is an evidence-based resource?
• Explicit methodology
• Comprehensive literature search
• Critically appraises evidence for validity
• Valid and relevant evidence summarized
• Results
• Quality

6
Clinical cases-1

• A 55 y.o. male, history of COPD and DVT, presents
  to the Emergency Room with SOB and left sided
  pleuritic CP for 3 days. Hospitalized for COPD
  exacerbation 3 wks earlier.
• On PE his pulse is 110 bpm, resp 22 bpm,
  scattered rhonchi on lung exam.
• No swelling or tenderness in legs.
• ECG shows sinus tachycardia. CXR shows a small
  infiltrate in left lower lung.
• WBC 13,000 with normal differential. Nl pAO2 on
  ABG.

7
Questions

• How likely is this patient to have a pulmonary
  embolus? (How do you determine this? Does it
  matter?)
• What diagnostic test (if any) would you use in
  this patient for pulmonary embolism?
• The patient undergoes CT angiography which is
  read as negative for PTE. What do you tell the
  patient when he asks if he has a PTE?

8
Clinical cases-2

• 62yo M referred for preop eval prior to THA. Pt
  has DM (oral agent), HTN and OA. Denies chest
  pain or DOE. Serum creatinine 1.6 mg/dl. BP
  normal. What is this patients cardiac risk for
  this surgery? Does he need any further testing?
• 70yo F with DM presents with 1 month h/o
  exertional substernal heaviness lasting 3-5 min.
  Relieved with rest. Smokes 1ppd. LDL 99 mg/dl. PE
  normal. ECG with no ischemic changes. Undergoes
  exercise SPECT which was negative. Is any
  further testing needed in this patient?

9
Clinical cases-3

• 55 yo F with DM controlled on rosiglitazone and
  glyburide comes in for opinion on continuing
  rosiglitazone. Cant take metformin due to nausea
  and creat 1.8mg/dl. Insurance doesnt cover
  pioglitazone. What do you tell her?

10
Clinical cases-4

• 82yo M with Parkinsons disease presents with 1
  day h/o of rigors, fever to 101 F, and cough
  productive of rust colored sputum. PE
  demonstrates consolidation in the right mid lung.
  CXR shows a RML infiltrate with air bronchograms.
  WBC is 14k with left shift. Does this patient
  have pneumonia?

11
(No Transcript)
12
Objectives

• Determine when which resources to use to answer
  different types of clinical questions
• Become familiar with evidence resources available
  at Lister Hill Library
• Understand the role of evidence in making
  clinical decisions

13
The Basic Skills of EBM

• Ask a clinically relevant question.
• DOE vs. POE
• Acquire the best available evidence.
• Appraise the quality importance of the
  evidence.
• Apply the evidence in patient care decisions.

14
Two types of clinical questions

• Background questions
• Asks about general information
• Foreground questions
• Asks about specific, patient-focused information
• PICO format

15
Choosing Resources
Foreground
Background
Unfiltered Database (e.g. MEDLINE)
Textbooks
Rare
Filtered/ Pre-appraised Evidence
Common
16
How to seek evidence-based information
Start at the top for every question!... Working
downward until you find the answer
Haynes, RB. ACPJC Nov/Dec 2006145(3)A8-9
17
How to seek evidence-based information
Computer system
Clinical Evidence or PIER (UpToDate)
ACP Journal Club, InfoPOEMS, Dynamed
Cochrane Library, PubMED Clinical Queries,
BMJUpdates, guidelines
Original Studies
OR SUMsearch or TRIP
18
Evidence needed for diagnostic testing

• Pre-test probability
• Operating characteristics of test(s)
• Sensitivity, specificity or likelihood ratios

19
Evidence needed for therapy

• Patients values/desires
• Pre-treatment probability of bad outcomes
• Benefits and harms of intervention(s)
• ARR and ARI
• NNT and NNH

20
How good is an evidence resource?

• Relevance
• Breadth of content?
• Depth of content?
• Applicability information?
• Validity
• Explicit methods?
• Updated?
• Reference? Linked?
• Evidence grading?
• Consistency?
• Perspective/bias?

• Accessibility
• Available? Cost?
• Platforms (web, desk-top, print, pda)
• Ease of Use
• Easy to log on use?
• Intuitive?
• Technically reliable?
• Search engine?

Adapted from Whelan, Ortiz, Shaneyfelt, Straus
for the SGIM EBM Task Force
21
Objectives

• Determine when which resources to use to answer
  different types of clinical questions
• Become familiar with evidence resources available
  at Lister Hill Library
• Understand the role of evidence in making
  clinical decisions

22

• Exploring Lister Hill Library EBM resources

23
UpToDate

• Not considered an EBM resource (by EBM experts)
• Electronic text book
• Limitations
• Not explicit (but claims to be)
• Inconsistent methodology
• Biases
• Incomplete evidence summary

24
So which product is best?

• Ranking of products by weighted importance
• ACP PIER
• Clinical Evidence
• Diseasedex- General Medicine
• DynaMed
• InfoPOEMS/InfoRetriever
• .
• UpToDate

Trumble et al. University of Texas Medical
Branch. Presented at SCC/MLA 2006
25
Objectives

• Determine when which resources to use to answer
  different types of clinical questions
• Become familiar with evidence resources available
  at Lister Hill Library
• Understand the role of evidence in making
  clinical decisions

26
EBM model for evidence based clinical decisions
Clinical expertise
ACPJC 2002 Mar-Apr136A11-14
27
Patients clinical state, the clinical setting,
and clinical circumstances
Clinical skills and experience (identify each
patients unique health state, diagnosis, risks
and benefits) Encompasses and balances other
components
Depends on their condition, values, experiences,
risk taking, insurance, family, willingness to
take meds, information, etc. Actions may differ
from preferences
Most valid and clinically relevant evidence for
a specific clinical decision Personalized
evidence
Clinical expertise
ACPJC 2002 Mar-Apr136A11-14
28
Summary

• Start at the top of the pyramid work down
• Not all resources are created equal
• Evaluate them before using them
• Use a couple of resources that complement each
  other (e.g. InfoRetriever DynaMed)
• Download to PDA
• Evidence is only 1 component of providing quality
  care