Transcript Evidence Based Healthcare 回想一下

Using Evidence from Clinical Trials to
Optimize Quality of Medical Care
李智雄醫師
高雄醫學大學附設醫院
實證醫學中心
實證醫學、循證醫學、證據醫學
Evidence Based Medicine
Evidence Based Dentistry
Evidence Based Pharmacy
Evidence Based Physiotherapy
Evidence Based Nursing
Evidence Based Nutrition
……………………………..
Evidence Based Healthcare
回想一下
過去的三個月中，針對病人的臨床問
題，您做過幾次完整的實證資料查詢與
評讀？
照顧病人真的有這麼多臨床問題嗎?
Resident’s information needs
• Setting: 64 residents at 2 New Haven hospitals
• Method: Interviewed after 401 consultations
• Questions
– Asked 280 questions (2 per 3 patients)
– Pursued an answer for 80 questions (29%)
– Not pursued because
• Lack of time
• Forgot the question
• Sources of answers
– Textbooks (31%), articles (21%), consultants (17%)
Green, Am J Med 2000
圖三 整體實習醫師第2題排名第一各選項百分比
實習醫師調查：遇到臨床問題的解決方式
其他, 1.4%
上課講義或教科
書, 13.5%
網際網路, 14.9%
上課講義或教科書
請教師長
同儕討論, 12.2%
同儕討論
網際網路
其他
請教師長, 58.1%
執行EBM的五個步驟 ( I )
1.問問題 ASK（可以回答的問題）
– Converting the need for information into an
answerable question.
2.找資料 ACQUIRE（可獲得最好的證據資訊）
– Search the database and tracking down the
best evidence.
3.分析判斷 APPRAISE（文獻的效度與重要性）
– Critical appraising that evidence for its
validity and importance.
執行EBM的五個步驟 ( II )
4.臨床應用 APPLY（整合三大層面）
– Integrating the critical appraising with our
clinical expertise and our patient’s unique
biology, values and circumstances.
5.評估成果 AUDIT（執行EBM的效率）
– Evaluating our effectiveness and efficiency
in executing step 1- 4 and seeking ways to
improve them both for next time.
Challenges
•
•
•
•
•
•
•
Too many patients
Too many problems
Too many tasks
Mental fatigue
Heaps of information
Staying in control
Maintaining the passion
圖十一 第17題「您認為要執行實證醫療決策最大的障礙是？」之圓餅圖分析
老師也不清楚無法指
導, 6.7%
沒有誘因, 2.5%
沒有興趣, 5.0%
電腦使用不方便,
3.4%
工作太忙時間不足,
27.7%
工作太忙時間不足
病人問題太多
文獻資料太多
找不到資料
電腦使用不方便
沒有興趣
找不到資料, 26.1%
病人問題太多, 10.1%
沒有誘因
老師也不清楚無法指導
文獻資料太多, 18.5%
Process of Evidence-Based Decision Making
• Fear of criticism
• Conflict with usual care
• Logistic constraint
• Cost
• Medicolegal concerns
Asking questions
Inertia gap
Clinical questions
Relevance gap
Applicable
evidence
Published relevant
research
Applicability gap
Retrievability gap
Retrieved relevant
evidence
Critical appraisal gap
Critically appraised
evidence
Dismantling the Barriers
• Attitudes of inquiring and asking questions
– Encourage questions during ward round
– Keep a question log book / PDA / handphone
• Information at the point of decision making
– Have the evidence sources at the point of care
• Lack of skills and knowledges of EBM
– Preappraised resources
– Rapid appraisal methods
• Lack of time
– Replace most passive learning with question-focused learning
– Use more effective updating methods
– EBM journal club (Clinical problem-oriented)
“Just in Time” learning
The EBM Alternative Approach
• Shift focus to current patient problems
(“just in time” education)
– Relevant to YOUR practice
– Memorable
– Up to date
• Learn to obtain best current answers
Dave Sackett
The "5S" levels of organisation of evidence from healthcare research
Brian Haynes, R Evid Based Med 2006;11:162-164
Evidence-based
CPGs
Copyright ©2006 BMJ Publishing Group Ltd.
The Quality of Health Care Delivered to Adults in the United States
N Engl J Med 2003;348:2635-45
•
Evaluate performance on 439 indicators of quality of care
–
–
–
•
•
October 1998 - August 2000
12 metropolitan areas in the United States
30 acute and chronic conditions, also preventive care
Senile cataract: 78.7% (95% CI 73.3 – 84.2)
Alcohol dependence: 10.5% (95% CI 6.8 - 14.6)
Number Needed to Search to Improve Care
• Random sample 146 inpatients cared for by 33
physicians
• Literature searches following formulation of
diagnosis and treatment plans, with feedback to
physicians
• Outcomes
– No. of patients for whom physicians improved
management due to searchs, as ascertained by blinded
peer review
• Results
– Plans changed in 18% (23/130) of eligible patients
– Peer reviewers judged quality of care to have been
improved or sustained in 78% (18/23) of treatment
changes
– NNS to improve care for 1 patient = 130/18 = 7 patients
Lucas et al J Gen Intern Med 2004
Getting Evidence into Practice
Evidence from clinical trials
Searching out or receiving high quality evidence
Apply high quality evidence in clinical decision making
Integration of evidence into practice
From Clinical Trials to Practice
Tools of Translation
•
•
•
•
Likelihood ratios for diagnostic tests
NNT and NNH for therapies
Clinical prediction rules
Clinical practice guidelines
Likelihood Ratios
sensitivity
1
LR (+) =
- sensitivity
LR (- ) =
1 - specificity
specificity
Pretest odds = prevalence / (1 – prevalence)
Pretest odds x LR = Posttest odds
Posttest Probability =posttest odds / posttest odds+1
Rule of 15%
LR+
10
5
2
LRor
or
or
0.1
0.2
0.5
Change in post-test probability
45%
30%
15%
The Likelihood Nomogram
Number Needed to Treat for Therapies
• NNT=Number needed to treat to prevent one
outcome ( 1 / ARR )
• Measure of the clinical impact of therapies
• Assists in choosing and prioritising treatment
options
• Preferable to use single common outcome
measures
• Should also concern about
–
–
–
–
Event
Treatment intensity / co-interventions
Duration of follow-up
Baseline patient risk
Clinical Prediction Rules
• Use of clinical findings to make a diagnosis or
predict an outcome
– History
– PE
– Test results
• Derived from systematic clinical observation
Purposes:
• Suggest a diagnostic or therapeutic course of action
• Change clinical behavior
• Reduce unnecessary costs
• Maintaining quality of care
Evaluate Pretest Probability
Low
Normal echo
No DVT
Moderate
Abnormal echo
No DVT
Abnormal echo
Repeat echo
in 3-7 days
venogram
-
Normal echo
+
+
DVT
High
No DVT
DVT
Normal echo
Abnormal echo
Venogram
DVT
+
DVT
No DVT
Clinical Practice Guidelines
• National Guideline Clearinghouse (NGC)
– http://www.guidelines.gov/
• New Zealand Guidelines Group
– http://www.nzgg.org.nz/index.cfm
• National Institute for Health and Clinical Excellence (NICE)
– http://www.nice.org.uk/
• Medical Information Network Distribution Service (Minds)
– http://minds.jcqhc.or.jp/index.aspx
• 國家衛生研究院 – 實證臨床指引平台
– http://ebpg.nhri.org.tw/
Check for
• Validity
• Grading of evidentiary strength of recommendation
• Accessibility
• Usability of format
• Applicability to local circumstances
Some Common Problems in Translating
Evidence from Clinical Trials to Practice
1. Generalizing Trial Results
Am Heart J 2003; 146:250-7
Heart Failure Trials
Trial Patients Community Patients
50 – 70 yrs
Mostly > 70 yrs
M>F
M~F
Diagnosis
Mainly CHF
Comorbidity
LV Function
Systolic
dysfunction
Systolic / diastolic
dysfunction
Treatment
Heart failure
Concomitant
Optimal
Variable
Age
Gender
Compliance
Spironolactone in Heart Failure
J Am Coll Cardiol 2003;41:211– 4
N Engl J Med 1999:341:709-17
Spironolactone prescription rate
(per 1000 patients)
Rate of in-hospital death from hyperkalemia
(per 1000 patients)
Rate of admission for hyperkalemia
(Per 1000 patients)
Rate of readmission for heart failure
Per 1000 patients
N Engl J Med 2004;351:543-51.
2. Faulty Comparators
Use of placebo when active comparator optimal
Prevention of diabetic nephropathy
Placebo
ARB
VS
ACEI
(N Engl J Med 2001;345:870-8.)
N Engl JMed 2001;345:861-9
RCT of High Dose Atorvastatin VS Moderate
Dose Pravastatin in ACS Patients
16% RRR at 2 years
N Engl J Med 2004;350:1495-504.
3. Surrogate End-Points
Results based on surrogate outcomes
Results based on clinical end points
Milrinone improved LV function during
exercise
Large RCT and meta-analysis showed
28% increase in mortality
Encainide suppressed VT in post-MI
patients
Large RCT showed 50% increase in
mortality
β-blockers cause decline in EF in post-MI
patients
RCTs show 32% decrease in mortality
in patients with heart failure
GP IIb/IIIa antagonist in AMI in the
absence of PCI improve coronary blood
flow and resolve ST elevation
RCT shows no mortality difference and
increased bleeding risk
Anticholinesterase inhibitors improve
scores on performance scales
RCT shows no difference in mortality,
carer burden, health care costs
4. Relative VS Absolute Measures of Benefit
對照組的風險
CER
實驗組的風險
EER
相對風險性降低度 絕對危險性降低度
RRR
ARR
70%
35%
50%
35%
7%
3.5%
50%
3.5%
0.7%
0.35%
50%
0.35%
• 相對風險性降低度 (RRR)無法呈現實際風險降低程度，亦沒有
考慮起始風險
• 絕對危險性降低度 (ARR)更準確表示治療效果，但亦不容易體
會兩組的差別
Number Needed to Treat (NNT)
“益一需治數＂
“益一需治數”：為了預防一個不良結
果
或減少一人死亡所需治療的病人數
例如：治療五人可減少一人死亡 VS 治療兩千人可減少一人死亡
NNT = 1 / ARR or 100 / ARR (%)
Framing Effect
• Physicians are more likely to prescribe
drugs when trial results are presented only
with information about RRR
“For those who are likely to be influenced by data
presentation, never, ever, accept information on the
basis of relative risk alone”
Misinformation
Level of Evidence for Class A and Class B Claims
Class A
Class B
(n=418)
(n=437)
Unreferenced claims
6 (1%)
References not on Medline
146 (35%)
Level 1 evidence (meta-analyses)
40 (10%)
Level 2 evidence (≧1RCT)
189 (45%)
Level 3 evidence
37 (9%)
58 (13%)
174 (40%)
59 (14%)
108 (25%)
38 (9%)
• 7.4% (13/174) reported quantitative statistics about outcomes
• 77% (10/13) reported RRR without additional information
• 8% (1/13) reported RRR with information allowing ARR and NNT calculation
• 15% (2/13) reported original data allowing RRR, ARR and NNT calculation
• No advertisement explicitly reported ARR or NNT
MJA 2002; 177:291-293
5. Use of Composite End-Points
In comparison to amlodipine, Irbesartan reduced the combined endpoint of all cause
mortality, progression to end stage renal disease, and doubling of serum creatinine
RRR 20%, 95% CI 7.5% - 32%
(N Engl J Med 2001;345:851-60.)
(N Engl J Med 2001;345:851-60.)
JAMA 2003; 289:2554-2559
Primary Composite Outcome and Mortality
From 1997 to 2000, review of 167 original reports of randomized
trials (with a total of 300276 patients) that included a composite
primary outcome that incorporated all-cause mortality
A high proportion of trials that measure composite outcomes, including
mortality, provide neutral results on the primary outcome may be unsurprising.
However, the finding that a similar proportion are positive yet fail
independently to identify an effect on the mortality component is striking and
requires further consideration
JAMA. 2003;289:2554-2559
Effects of Clinician-driven End-Points
• 78 of 179 comparisons (including 20 primary
outcomes from studies with multiple comparisons)
included the following clinician-driven outcomes:
– revascularization, percutaneous mitral valvuloplasty,
mechanical ventilation, hospitalization, transplantation,
use of rescue therapy, initiation of new antibiotics, use of
shock therapy, amputation, ECMO, dialysis etc.
• The inclusion of a clinician-driven outcome was
predictive of a statistically significant result for the
primary composite outcome
OR 2.24 (95% CI 1.15-4.34); P =0.02
JAMA. 2003;289:2554-2559
6. Small Effect Size
N Engl J Med 2003;348:583-92
7. Sponsor Bias
May 2001 Cochrane Library, 167 Cochrane reviews
6 Points: Experimental intervention highly preferred and should now be considered
the standard intervention in all patients, or similar
JAMA. 2003;290:921-928
Independent Predictors for Stronger Recommendation
JAMA. 2003;290:921-928
JAMA. 2006;295:2270-2274
JAMA. 2006;295:2270-2274
Recommendation
• Write down all your clinical questions
• Be familiar with the search strategy and
database available, especially preappraised resources
• Just in time learning
• Understand the pitfalls of using clinical trial
results