The Berlin Definition: Why Necessary?

The Berlin Definition: Why Necessary? Gordon D. Rubenfeld, MD MSc Professor of Medicine, University of Toronto Chief, Program in Trauma, Emergency, a...
Author: Crystal Hoover
5 downloads 0 Views 1MB Size
The Berlin Definition: Why Necessary?

Gordon D. Rubenfeld, MD MSc Professor of Medicine, University of Toronto Chief, Program in Trauma, Emergency, and Critical Care Sunnybrook Health Sciences Centre

Outline • What Psychiatry and Critical Care have in common • Everything we know about the ARDS clinical phenotype we learned in 1967 • The Berlin Definition and what it changed, didn’t change, what it will fix, and what it won’t fix

Syndrome definitions get updated and its always controversial and political

And the criticism is similar (except we have objective lab measures) This volume will tweak several current diagnostic categories, from autism spectrum disorders to mood disorders. While many of these changes have been contentious, the final product involves mostly modest alterations … The weakness is its lack of validity. Unlike our definitions of ischemic heart disease, lymphoma, or AIDS, the DSM diagnoses are based on a consensus about clusters of clinical symptoms, not any objective laboratory measure.

Tom Insel, MD Director NIMH

Old and New Testament

1994

2013

AECC definition – Reliability What’s wrong with the Old Testament?

• • • • •

Experts disagree on radiograph (= 0.5) Poor inter-observer agreement on PAOP P/F ratio is treatment dependent (TV, PEEP, FiO2) Left atrial hypertension often co-exists with ALI Reliability of clinical assessment of risk factors, direct vs indirect injury, and “clinical exclusion of LAH” completely unstudied • ALI term used incorrectly Rubenfeld,, Chest. 1999;116:1347-1353. Ferguson ND, Intensive Care Med 2002; 28(8): 1073-7. Villar J, Intensive Care Med 1999; 25(9): 930-5.

AECC definition – Criterion validity What is the gold standard for ARDS? 140 120

Cases

100 80 60 40

DAD

Only 75% of AECC cases of ARDS show DAD

20 0 AECC

Esteban, et al Ann Intern Med. 2004;141:440-445

If the pathologic gold standard is DAD – not great If it includes edema and pneumonia – not bad 140 Edema

100

Pneumonia

Cases

120

80 60 40

DAD

20

97% of AECC cases of ARDS with DAD, pneumonia, or edema

0 AECC

Esteban, et al Ann Intern Med. 2004;141:440-445

Why the Berlin Definition can’t change too much or … everything we know about the phenotype we learned in 1967

Acute onset, hypoxemia, low compliance, chest radiographic appearance

Risk factors

PEEP responsiveness

Pathology

Everyone agrees we need a better ARDS definition but, what does better mean? • Reliability, Validity, Feasibility ⦁ How do you study validity without a gold standard?

Potential impact of recent changes in the definition of ARDS for clinical research • The Berlin ARDS definition is evolutionary not revolutionary • Main changes: – Introduces the correct framework for thinking about and evaluating definition changes – Eliminates ALI term as it was being misused (ALI/ARDS) – 3 simple categories (mild, moderate, severe) – Requires minimal PEEP level for assessing gas exchange – Allows for diagnosis in non-intubated patients (mild) – Objectively evaluated (and rejected) a more complex severe definition – Provides educational materials to enhance reliability

Educational materials

Evaluation of Berlin Severe Ancillary variables identify a smaller group of patients with similar mortality

45% Mortality Severe 14%

Severe 28% Mild 22%

Mild 22%

Moderate 64%

Draft ARDS PaO2/FiO2 < 100 mmHg

PEEP > 10 cm H2O 3 or 4 quadrant opacities on CXR Crs < 40 ml/cm H2O VeCorr > 10 L/min

Moderate 50%

Mortality 45%

Final ARDS PaO2/FiO2 < 100 mmHg PEEP > 5 cm H2O Bilateral opacities on CXR

Evaluation of Berlin Severe Ancillary variables identify a smaller group of patients with similar mortality

Same Mortality ≠ Same Disease Same Mortality ≠ Same Pathophysiology Same Mortality ≠ Respond to Same Treatments There may be other reasons to include these variables

Comparative predictive validity (neither very good - these are not prediction models)

Berlin 0.58 AECC 0.54 P < 0.001 .

• • • • •

Does or does not add to predictive validity for mortality Does not correlate with DAD Does correlate with EVLW Generally critical of the exercise Generally missed the point of objectively evaluating proposed modifications, framework, and training

Concerns with Berlin Definition • Association with mortality poor – could have been a better prognostic model

• Not a prognostic model. The purpose of the empiric analysis was to asess the proposed severe definition • Association with mortality used to validate

Concerns with Berlin Definition • Association with mortality poor • Did not standardize mechanical ventilation at time of oxygenation assessment

• Oxygenation after standardized mechanical ventilation clearly associated with mortality better than baseline oxygenation

Day 1

Concerns with Berlin Definition • Association with mortality poor • Did not standardize mechanical ventilation at time of oxygenation assessment

• Oxygenation after standardized mechanical ventilation clearly associated with mortality better than baseline oxygenation • Requiring would eliminate all observational research and enrolment in trials requiring early intervention

Concerns with Berlin Definition • Association with mortality poor • Did not standardize mechanical ventilation at time of oxygenation assessment • Did not include: – CT, EVLW, PET, biomarkers …

• Test performance of these measures not evaluated in broad populations of patients with respiratory failure • Literature review on reliability and validity not persuasive • Not feasible for all centers • Berlin does not preclude using any for a specific trial

Moving the field forward? Berlin is just a step towards the “cancer” goal: mechanism guided therapy

• The “advances” of Berlin were methodological – Use of reliability, feasibility, and validity framework – A priori empiric evaluation methods – Beyond consensus

• Actual value in reliability should be tested

Email [email protected] for slides

Suggest Documents