E B M Must Not Be

If it is going to make any sense, EBM must not be… Adam La Caze University of Sydney

Background ’ EBM seems to have achieved some level of orthodoxy

’ Despite this there is intense, often polemic, debate in the medical literature regarding the validity of its claims1

’ Central to the debate is the notion of a ‘hierarchy of evidence’

’ Both sides agree that the philosophical suppositions of EBM are yet to be elucidated2

Most quoted definition: ’the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients

Sackett BMJ 1996 The evidence hierarchy Preliminary comments ’ Noting the original dialectic is useful: EBM vs Eminence-based medicine

’ Clarity is required on the motivating question: ’ What is it that advocates of EBM advocate? ’ What effect does the dictates of EBM have on practice? ’ What are the philosophical presuppositions of EBM? Are they defendable?

(I will argue) EBM can not be..

  1. Primarily about anything other than therapeutic decisions Need to argue for this based on methodology and philosophy of experiment

  2. A rigid hierarchy with randomised controlled trials always at the peak

Therapeutic Decisions ’ The population therapeutic decision: Are the benefits of a given therapy are likely to outweigh its harms in a given population of patients?

’ The individual therapeutic decision: Are the benefits of a given therapy likely to outweigh the harms in an individual patient?

The positive argument 1. All things being equal there is a defendable hierarchy of evidence for therapeutic decisions This still needs to be elucidated: {[red ?keep the current hierarchy or develop a new hierarchy; utilize philosophy of experiment (?other options within epistemology/phil of science) ]}

  1. That the..—all things being equal..—clause occurs in only very narrowly defined situations

  2. It is because of (2) that the hierarchy cannot be rigid and there is always a role for a range of evidence in therapeutic decisions..—but this is a debate occurring within the framework of (1) Need to find a way to express this more clearly

When are all things equal? ’when the only difference to the empirical evidence in response to a therapeutic question is the methodology (RCT, cohort, case-control)

’ Same population of patients, clinical outcomes, study duration, ect.

Evidential Hierarchy ’ Clearly this sees the hierarchy as an idealisation—but it is a defendable idealisation

’ This highlights just how limited the domain is in which the hierarchy will hold ’ In reality, all things are never equal; populations, clinical outcomes, study duration always differ between studies

’ There will always be a debate; but a debate within the framework of an idealised evidential hierarchy (I will call this debate EBM+)

Case 1: Does hormone replacement therapy effect cardiac risk? ’ Observational data ’ Data on over 100,000 women suggests that women who take HRT have a 35-50% lower risk of coronary heart disease1

’ Experimental data ’ Randomised controlled trial data on 16,000 women suggests that women who take HRT have a 21% higher risk of coronary heart disease2

’ Outcome: RCT trumps observation data ’ Media melee asking how doctor’s could get it so wrong

Case 1: Does hormone replacement therapy effect cardiac risk? 1. Attempts are made to understand the discrepancy ’ Differences in study population, age, reason for HRT use, potential confounders: life-long socioeconomic status

  1. Therapeutic decisions are made on the basis of these debates ’ Lack of clear role for long-term HRT in heart disease prevention; Maintenance of short-term role in menopausal symptoms

Case 2: Clopidogrel in combination with aspirin after a heart attack ’ Basic science rationale: both drugs inhibit platelets in different ways; the more ways you stop blood clots the better

’ RCT evidence: In patients with a certain type of heart attack the combination of aspirin and clopidogrel reduces a combined endpoint (CV death, non-fatal MI and stroke)1

’ Observational data: The use of aspirin and clopidogrel in combination is increasing. Odds ratio’s for serious upper GI bleeds: ASA alone 1.8; Clopidogrel alone 1.1; ASA + Clop 7.42

Case 2: Clopidogrel in combination with aspirin after a heart attack ’ Clinical use of the clopidogrel + aspirin combination has extended the select group of patients shown to benefit in RCTs

’ The observational data highlights some of the bleeding risks in these—closer to ‘real-life’—patients

’ The observational data enriches and constrains the RCT findings and application

EBM+ is not about questions other than therapeutic ’ Different questions need different research methods ’ Including disease aetiology and prognosis, public health, sociology and government policy

’ EBM+ does not provide a research (or research funding) agenda

’ EBM+ does not deny the importance of basic science and clinical experience in forming research questions and designing trials

The EBM+ hierarchy can not be rigid 1. The hierarchy is only defendable as an idealisation ’ How should conflicting evidence from different tiers of evidence be understood?

  1. The debate/application of EBM+ requires knowledge of observational studies, basic sciences and clinical experience ’ Can the results be extrapolated to this patient?

Discussion ’ If EBM+ adequately outlines the philosophical nub of EBM it provides a partitioning of EBM counter-arguments

’ In the context of the EBM debate in the medical literature ’ Highlighting the fallibility of RCTs is not an argument against EBM But ’ EBM rhetoric on the benefits of RCTs needs to be curtailed

Some replies.. And more questions ’ What is it that advocates of EBM advocate? ’ Can the idea that EBM is actually EBM+ be defended?

’ What effect does the dictates of EBM have on practice? ’ Here is the biggest area of problems for EBM ’ Asymmetry of EBM; Role of industry funding; Problematic extension of EBM to other areas; Problem of the role of classical statistics in EBM

’ What are the philosophical presuppositions of EBM? Are they defendable? ’ Little, to no, debate on the assumptions of EBM+ in the literature ’ Are they on as sure a footing as is assumed? This needs much discussion’ ?separate paper

Some paper ideas: ’ Stucture of EBM paper: EBM+ and the sense it can make of some of the claims of EBM. Assume evidence hierarchy and remain explicitly agnostic about whether EBM+ is what EBM advocates really mean (cf rhetoric with what actually occurs in practice)‘Provide a proposal for the philosophical core of EBM (perhaps I need to explicitly outline a philosophy of experiment for this). Use cases..—HRT, clopidogrel+asa’ ?more ’ Epistemology of EBM+: can it be defended? How? Philosophy of experiment/options?