can addiction research be trusted?

59
Can Addiction Research Can Addiction Research Be Trusted? Be Trusted? Stewart B. Leavitt, PhD, Editor Addiction Treatment Forum Sponsored by an educational grant from Mallinckrodt Inc. Introducing EBAM* (*Evidence Based Addiction Medicine)

Upload: medresearch

Post on 21-Jun-2015

199 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Can Addiction Research Be Trusted?

Can Addiction Research Can Addiction Research Be Trusted?Be Trusted?

Stewart B. Leavitt, PhD, EditorAddiction Treatment Forum

Sponsored by an educational grant from Mallinckrodt Inc.

Introducing EBAM*(*Evidence Based Addiction Medicine)

Page 2: Can Addiction Research Be Trusted?

2

EBAM Guides AvailableEBAM Guides Available “Can Addiction Research Be Trusted?” – report Available online at www.atforum.com

Look under Addiction Resources tabSlides from this presentation also available

Called “EBAM AATOD Slides” PowerPoint format

Page 3: Can Addiction Research Be Trusted?

3

““EBAM for Practitioners”EBAM for Practitioners”Available at…

www.atforum.com under “Addiction Resources” tab.

A more technical explanation for persons who need to critically assess research evidence.

Page 4: Can Addiction Research Be Trusted?

4

Can Addiction ResearchCan Addiction ResearchBe Trusted?Be Trusted?

Yes? No? Sometimes?Yes? No? Sometimes? Research must be critically examined

and interpreted for particular clinical needs and patient benefits.

Just because something appears in print does not mean it is worthwhile or valid.

Page 5: Can Addiction Research Be Trusted?

5

Why Should You Care?Why Should You Care?

Research directly or indirectly affects the practice of addiction medicine on a daily basis.

There is an urgent need for good science to overcome stigma, prejudice, and misunderstanding.

Trepanation once was a “scientific cure” for

mental illness.

Page 6: Can Addiction Research Be Trusted?

6

What is EBAM?What is EBAM?Evidence-Based Addiction Medicine Evidence-Based Addiction Medicine

The use of current best evidence in making decisions about the care of patients.

Involves combining clinical expertise with valid research on a topic of concern.

Differentiates between practices based on sound evidence versus those founded more on 1) tradition, 2) outdated beliefs, or 3) “junk” science.

Page 7: Can Addiction Research Be Trusted?

7

Why EBAM?Why EBAM?

EBAM approaches allow you to respond more convincingly to the following questions…. Where did you hear about that treatment? How do you know the information is worthwhile

and true? What do you propose doing and what results do

you expect? What are the costs versus benefits?

Page 8: Can Addiction Research Be Trusted?

8

Essential UnderstandingsEssential Understandings

Medical research is an imperfect science.

Medical writing/reporting is often biased.

Scientific research does not “prove” anything to be true or not true.

Page 9: Can Addiction Research Be Trusted?

9

The Nature of Clinical ResearchThe Nature of Clinical ResearchCompares experimental treatment (intervention)

with standard therapy or placebo.Research subjects represent only a tiny portion of

the population in question – statistical analyses estimate effects in the larger population.

All things are possible (some patients recover without treatment), and chance can play an important role in the results.

Page 10: Can Addiction Research Be Trusted?

10

Interpreting ResearchInterpreting Research Interpretation is a matter of

perception. Not everyone “sees” a

research study in the same way.

Becoming a more critical consumer of research is essential.

Page 11: Can Addiction Research Be Trusted?

11

Problems with Medical Problems with Medical Research Articles OverallResearch Articles Overall

Surveys of research have found startling results… On average, half or more of journal articles contain

errors – from missing data to major design flaws. Roughly, up to 2/3 of article abstracts may contain

data that is either inconsistent with or missing from the main body of the article.

In one investigation, 80% of systematic reviews & meta-analyses examined had serious flaws.

Page 12: Can Addiction Research Be Trusted?

12

Concerns About Addiction Concerns About Addiction Research StudiesResearch Studies

Many articles appear to be written for other researchers, rather than practitioners.

There are often too few subjects and/or inadequate effects for accurately detecting significant group differences.

Subjects are often selected to maximize outcome results, but they are not typical patients.

Cases are presented as if they are evidence of causality or applicable to broader groups.

There may be a biased leap of interpretation from laboratory experiments to potential effects in humans.

Page 13: Can Addiction Research Be Trusted?

13

What Really Matters?What Really Matters?

“Not everything that can be counted counts, and not everything that counts can be counted. What really mattersis not always obvious.” – Albert Einstein

Just because data were collected, analyzed, and reported does not mean they are relevant to the clinical question at hand – but it mattered to somebody.

Page 14: Can Addiction Research Be Trusted?

14

What Are What Are Significant Significant Outcomes?Outcomes?

Statistical Significance – the likelihood (probability) that the results represent true differences between groups.

Clinical Significance – importance of the results for providing meaningful benefits to patients.

Statistical significance does not automatically transfer to clinical significance.

Page 15: Can Addiction Research Be Trusted?

Taking A Closer LookTaking A Closer LookAtAt

Clinical Research ArticlesClinical Research Articles

Page 16: Can Addiction Research Be Trusted?

16

Anatomy of a Research ArticleAnatomy of a Research Article

AbstractIntroduction

MethodsResults

DiscussionConclusion

Page 17: Can Addiction Research Be Trusted?

17

AbstractAbstractPresents overview: who, what, where,

when, why, and how.Structured abstract is a mini-article…

Summarizes major sections of article. Limited to 250 words.

Abstract may be inaccurate & misleading.

Page 18: Can Addiction Research Be Trusted?

18

IntroductionIntroductionStates purpose of study, including:

clinical background, review of previous research, rationale for current investigation.

Identifies questions (hypotheses) to be investigated in present study.

Page 19: Can Addiction Research Be Trusted?

19

MethodsMethodsDefines the patients (subjects) included in

the study and the sample sizes.Describes in detail how the study was

conducted (treatment protocol).Defines outcomes of importance and how

they were measured.Describes how data was analyzed (statistics).

Page 20: Can Addiction Research Be Trusted?

20

ResultsResultsDescribes in detail what was found:

differences in treatment effects (outcomes), their significance (p values), and, confidence limits.

Should show how statistical procedures were performed and any data adjustments (eg, for noncompliant/dropout subjects).

Page 21: Can Addiction Research Be Trusted?

21

DiscussionDiscussion

Recaps results (the “payoff”).Interprets findings in comparison

with other research on the topic.Describes study limitations.Notes other factors that might have

influenced outcomes.

Page 22: Can Addiction Research Be Trusted?

22

ConclusionConclusion

Presents case for “external validity” – how, when, where, and in which patients the results might be applied in clinical settings.

Often provides suggestions for further research on the subject.

Page 23: Can Addiction Research Be Trusted?

23

Hierarchy of EvidenceHierarchy of EvidenceClinical Therapy ResearchClinical Therapy Research

Systematic Reviews / Meta-Analyses of RCTs.

Randomized Controlled Trials (RCTs).

Cohort Studies

Case-Control Studies

Cross-Sectional Surveys

Case Reports

“Perspectives” Articles

Page 24: Can Addiction Research Be Trusted?

24

““Perspectives” ArticlesPerspectives” ArticlesAka* overviews, reports, commentary,

editorials, or interviews. *Also known as…

Most common in addiction field; often cited as evidence.

Summarize or comment on research, and are the most subject to bias.

Usually, favors one viewpoint (perspective) over another.

Page 25: Can Addiction Research Be Trusted?

25

““The Emperor’s NewThe Emperor’s NewPerspective (Clothes)”Perspective (Clothes)”

“He thought it better to continue the illusion that anyone who couldn’t see his clothes was either stupid or incompetent.” – H.C. Anderson

Sometimes, the higher the “authority” of the source, the less its perspective can be trusted as being valid and unbiased.

Page 26: Can Addiction Research Be Trusted?

26

Case ReportsCase Reports

Aka case histories, case series, or anecdotes.Draw upon personal observations or medical

records to report unexpected events.Subject to errors in observation or interpretation,

or inadequate documentation.May arrive at unsupported conclusions.“The plural of anecdote is not evidence.” – Leshner

Page 27: Can Addiction Research Be Trusted?

27

Cross-Sectional SurveysCross-Sectional Surveys

Aka prevalence or epidemiology studies.Examines relationships of medical conditions,

treatments, and/or other factors in a patient population at a particular time.

Can establish associations but not causality. May be problems with recall and extraneous

factors (confounders).

Page 28: Can Addiction Research Be Trusted?

28

Case-Control StudiesCase-Control StudiesAka case-referent, case-comparison, or

retrospective studies.Identifies patients with outcome of interest (Cases)

and Control patients without the outcome.Researchers look back in time to see how the two

groups differed (especially, regarding treatment or exposure received).

Subject to recall bias or inconsistent records.

Page 29: Can Addiction Research Be Trusted?

29

A “Classic” Case-Control StudyA “Classic” Case-Control Study25 persons (Cases) exposed to

Tutankhamen’s tomb in 1923-26.Compared with 11 explorers in Egypt

at that time not exposed (Controls).

Results: No significant differences in mean age at death or in years survival after exposure.

Conclusion: Mummy’s curse was an urban legend.

Page 30: Can Addiction Research Be Trusted?

30

Cohort StudiesCohort StudiesAka followup, incidence, longitudinal,

observational, or prospective studies.Most common trial in addiction medicine.An Experimental group (cohort) receives

treatment of interest; Control group does not.Outcome differences between groups are

compared and contrasted.Nonrandomized, many factors uncontrolled.

Page 31: Can Addiction Research Be Trusted?

31

A First A First (?) (?) Clinical Cohort TrialClinical Cohort Trial

In 1774, a ship’s doctor took 12 sailors with scurvy and treated them daily, 2 at a time, with either……cider, elixir vitriol, vinegar, sea

water, nutmeg, or oranges/lemons.Compared with untreated patients.

Citrus fruit provided rapid relief.

Page 32: Can Addiction Research Be Trusted?

32

Randomized Controlled TrialsRandomized Controlled Trials

RCTs considered “gold standard.”Subjects randomized to Experimental and Control

conditions; then followed for outcomes.Extraneous factors (confounders) are equally

distributed across groups.Subjects/investigators usually “blinded.”May be ethical concerns about Control condition.

Page 33: Can Addiction Research Be Trusted?

33

Systematic ReviewsSystematic Reviews& Meta-Analyses& Meta-Analyses

Gather highest quality evidence on a subject.Compare and contrast studies to reach summary

statements and conclusions.Meta-analyses combine statistical evidence from

multiple studies for greater precision.Conclusions tend to be reliable and accurate.Clinical guidelines often come from these studies.

Page 34: Can Addiction Research Be Trusted?

A Closer LookA Closer LookAtAt

Clinical Research ValidityClinical Research Validity

Page 35: Can Addiction Research Be Trusted?

35

What is Validity?What is Validity?The degree to which the research represents the

most accurate and bias-free treatment effects.Internal Validity – refers to effects that are valid

for the subjects in the particular study. External Validity – refers to the relevance of a

study for typical addiction treatment settings.Validity is determined by looking at a study’s

design and sources of potential bias.

Page 36: Can Addiction Research Be Trusted?

36

What is Bias?What is Bias?

Bias is anything that influences conclusions about the groups under investigation and potentially distorts comparisons.

The question is not if there is bias, but how much and whether it is sufficient to make the study invalid for a particular purpose.

Page 37: Can Addiction Research Be Trusted?

37

Sources of Potential Bias - 1Sources of Potential Bias - 1Publication Bias – Trials with “positive”

results are more likely to be submitted and published more quickly.

Patient Selection – Who was excluded or included? Were groups equivalent?

Subject Management – Were groups treated exactly the same, except for the intervention under study?

Page 38: Can Addiction Research Be Trusted?

38

Sources of Potential Bias - 2Sources of Potential Bias - 2

Randomization – Subjects should have an equal chance of assignment to each study condition.

Blinding (Masking) – Avoids preconceived ideas about treatment effects.

Placebo Effects – Mere expectation of benefit (or harm) may influence outcomes.

Page 39: Can Addiction Research Be Trusted?

39

Sources of Potential Bias - 3Sources of Potential Bias - 3Confounding Factors – Extraneous factors,

known or unknown, may cause certain effects to occur or not to occur.

Compliance/Follow-up – Adherence to protocol and retention in study are critical.

Completeness of Data – Are all or only some subjects analyzed? (ITT vs Per Protocol)

Page 40: Can Addiction Research Be Trusted?

40

Overcoming “Statistophobia”Overcoming “Statistophobia”

When most people hear the word statistics they want to hide.

It is not necessary to be a biostatistician to understand and assess addiction research.

Page 41: Can Addiction Research Be Trusted?

41

Example:Example: How Large Is The How Large Is TheTreatment Benefit?Treatment Benefit?

RCT comparing 8 mg/d buprenorphine (BUP) vs 80 mg/d methadone (M80).

Average illicit-opioid+ urinalyses at 26 weeks.

M80 produced significant 17% reduction. (55% - 38% = 17%; p < .01) Ling et al. Arch Gen Psychiatry.

1996;53:401-407.

Page 42: Can Addiction Research Be Trusted?

42

Example:Example: How Many Patients How Many Patients Need To Be Treated?Need To Be Treated?

Treatment effect = 17% reduction with M80.For every 100 patients treated with M80,

rather than BUP, 17 more would benefit.That is…

…for every 6 patients (100 ÷ 17) treated with M80, one additional patient would remain illicit-opioid abstinent than if treated with BUP.

Page 43: Can Addiction Research Be Trusted?

43

What The Article What The Article Didn’t Didn’t SaySay

In the 8 mg/d BUP vs 80 mg/d Methadone trial – The patients selected were ideal; rather than typical. There was a 95% chance that the “true” treatment

effect was between 7% and 28% (vs 17% reported). Not enough subjects enrolled (n=150) to accurately

detect the relatively small treatment effect of 17%. So many subjects dropped out that the validity should

be questioned. (48% M80, 65% BUP gone at 26 wks)

Page 44: Can Addiction Research Be Trusted?

Why DoesWhy DoesGood ResearchGood Research

Go Bad?Go Bad?

Page 45: Can Addiction Research Be Trusted?

45

Researcher-WritersResearcher-WritersFace ChallengesFace Challenges

Insufficient funding – researchers may choose to do “something” when funds are not available to do it “right.”

Inadequate training in research & reporting. Lack of technical support – many

researchers do not have access to data processing and statistical support staff.

Page 46: Can Addiction Research Be Trusted?

46

“Publish or perish” pressures – influence poor quality in design and reporting.

Political agendas – promote certain lines of research looking for “desired” results.

Journal space limitations – foster inadequate descriptions of analyses and results.

“Too many cooks” – multiple authors and peer reviewers complicate the process.

Researcher-WritersResearcher-WritersFace Challenges Face Challenges (Cont.)(Cont.)

Page 47: Can Addiction Research Be Trusted?

47

Research Writing Run AmokResearch Writing Run Amok

When too many people are involved in a research article, it reminds us that…

“A camel is a horse that was designed by a committee.”

Page 48: Can Addiction Research Be Trusted?

48

Dangers of “Data Dredging”Dangers of “Data Dredging”

Searching for “significant results” after the study is over and the findings were not as expected or desired (post hoc analysis).

Analyzing the same data in different ways until something “important” turns up.

Cranking-out several articles from a single small clinical study.

Page 49: Can Addiction Research Be Trusted?

49

Laboratory vs Clinical ResearchLaboratory vs Clinical Research

Test tube & animal results do not necessarily apply to people.

“Bench work” can be invaluable for guiding clinical trials in humans.

But the ultimate test is in patients “at the bedside.”

©1980, David Sinclair, PhD - used by permission

Page 50: Can Addiction Research Be Trusted?

50

Distortions in the Mass MediaDistortions in the Mass Media“Medical Propaganda”“Medical Propaganda”

Medical journalists usually lack understanding of research methods and shortcomings.

News editors seek sensational health stories to grab the public’s attention.

Research organizations and publications should be more cautious in their reporting of research to the press.

Readers need to become more critical.

Page 51: Can Addiction Research Be Trusted?

PuttingPuttingAddiction ResearchAddiction Research

Into PracticeInto Practice

Page 52: Can Addiction Research Be Trusted?

52

Judging Everyday RelevanceJudging Everyday Relevance

Is the article of high quality and valid?Does the study relate to your clinical needs?Are the participants similar to your own patients?Is the approach used in the study clearly

explained and free of bias?Are the results understandable and significant?Do the conclusions make sense and suggest clear

benefits for patient care?

Page 53: Can Addiction Research Be Trusted?

53

Develop A Healthy SkepticismDevelop A Healthy Skepticism

Learn to trust your educated instincts.

If it sounds like a duck, and walks like a duck, but smells like a fish… …something is probably a bit “fishy.”

Page 54: Can Addiction Research Be Trusted?

54

What Else Can You Do?What Else Can You Do?1. Only support/subscribe to publications

offering high-quality, well-documented research evidence.

2. Lobby NIDA & other funding organizations to provide clinically-relevant research of the best quality in design and reporting.

3. Challenge “fishy” reports in the media or from other sources – write letters.

Page 55: Can Addiction Research Be Trusted?

55

““Bible” for Medical AuthorsBible” for Medical Authors

AMA style book is the “gold standard” for medical writing.

Depicts correct formats for research papers, plus a wealth of other info.

Differs from other styles used in behavioral and social sciences.

Page 56: Can Addiction Research Be Trusted?

56

E-BM GuidebookE-BM GuidebookUsers’ Guides to the

Medical Literature (Textbook with CD-ROM)

Available at: www.usersguides.org www.amazon.com

Presents all aspects of evidence-based medicine.

Page 57: Can Addiction Research Be Trusted?

57

““EBAM for Practitioners”EBAM for Practitioners”Available at…

www.atforum.com under “Addiction Resources” tab.

A more technical explanation for persons who need to critically assess research evidence.

Page 58: Can Addiction Research Be Trusted?

58

Can Addiction Research Be Trusted?Can Addiction Research Be Trusted?

Usually!Usually!IF IF you learn what toyou learn what to

look for.look for.

“We have met the enemy and he is us!” - Pogo (Walt Kelly)

Page 59: Can Addiction Research Be Trusted?

59

Sponsored by an educational grant from Mallinckrodt Inc.