Evidence-based Dentistry
Nalaka Jayaratne BDS, PhD
Resident in Orthodontics,
University of Connecticut School of Dental Medicine
USA

Evidence-based Dentistry

Evidence-based Dentistry
 ADA Definition - an approach to oral health care that requires
the judicious integration of systematic assessments of clinically
relevant scientific evidence, relating to the patient's oral and
medical condition and history, with the dentist's clinical
expertise and the patient's treatment needs and preference

Why is evidence-based dentistry
important?

Why is evidence-based dentistry
important?
 To improve quality of health-care delivery

Why is evidence-based dentistry
important?
 To improve quality of health-care delivery
 By incorporating effective practices

Why is evidence-based dentistry
important?
 To improve quality of health-care delivery
 By incorporating effective practices
 While eliminating those that are ineffective or inappropriate

The five steps of evidence-based
dentistry practice

The five steps of evidence-based
dentistry practice
1. Developing a clear, clinically focused
question

The five steps of evidence-based
dentistry practice
1. Developing a clear, clinically focused
question
2. Identifying, summarizing, and
synthesizing all relevant studies that
directly answer the formulated question

The five steps of evidence-based
dentistry practice
1. Developing a clear, clinically focused
question
2. Identifying, summarizing, and
synthesizing all relevant studies that
directly answer the formulated question
3. Appraising evidence in terms of validity
and applicability

The five steps of evidence-based
dentistry practice
1. Developing a clear, clinically focused
question
2. Identifying, summarizing, and
synthesizing all relevant studies that
directly answer the formulated question
3. Appraising evidence in terms of validity
and applicability
4. Combining research evidence with
clinical expertise and patients
characteristics

The five steps of evidence-based
dentistry practice
1. Developing a clear, clinically focused
question
2. Identifying, summarizing, and
synthesizing all relevant studies that
directly answer the formulated question
3. Appraising evidence in terms of validity
and applicability
4. Combining research evidence with
clinical expertise and patients
characteristics
5. Assessing the successful
implementation of previous steps

Level Evidence
1++ High-quality meta-analyses, systematic reviews of RCTs, or RCTs with a very low risk of bias
1+ Well-conducted meta-analyses, systematic reviews of RCTs, or RCTs with a low risk of bias
1- Meta-analyses, systematic reviews or RCTs, or RCTs with a high risk of bias
2++
High-quality systematic reviews of case-control or cohort studies or high-quality case-control or
cohort studies with a very low risk of confounding, bias, or chance and a high probability that
the relationship is causal
2+
Well-conducted case-control or cohort studies with a low risk of confounding, bias, or chance
and a moderate probability that the relationship is causal
2-
Case-control or cohort studies with a high risk of confounding, bias, or chance and a significant
risk that the relationship is not causal
3 Non-analytic studies, e.g., case reports, case series
4 Expert opinion
Scottish Intercollegiate Guidelines Network (SIGN)
Revised grading system for recommendations in
evidence-based guidelines

Level Evidence
1++ High-quality meta-analyses, systematic reviews of RCTs, or RCTs with a very low risk of bias
1+ Well-conducted meta-analyses, systematic reviews of RCTs, or RCTs with a low risk of bias
1- Meta-analyses, systematic reviews or RCTs, or RCTs with a high risk of bias
2++
High-quality systematic reviews of case-control or cohort studies or high-quality case-control or
cohort studies with a very low risk of confounding, bias, or chance and a high probability that
the relationship is causal
2+
Well-conducted case-control or cohort studies with a low risk of confounding, bias, or chance
and a moderate probability that the relationship is causal
2-
Case-control or cohort studies with a high risk of confounding, bias, or chance and a significant
risk that the relationship is not causal
3 Non-analytic studies, e.g., case reports, case series
4 Expert opinion
Scottish Intercollegiate Guidelines Network (SIGN)
Revised grading system for recommendations in
evidence-based guidelines

Level Evidence
1++ High-quality meta-analyses, systematic reviews of RCTs, or RCTs with a very low risk of bias
1+ Well-conducted meta-analyses, systematic reviews of RCTs, or RCTs with a low risk of bias
1- Meta-analyses, systematic reviews or RCTs, or RCTs with a high risk of bias
2++
High-quality systematic reviews of case-control or cohort studies or high-quality case-control or
cohort studies with a very low risk of confounding, bias, or chance and a high probability that
the relationship is causal
2+
Well-conducted case-control or cohort studies with a low risk of confounding, bias, or chance
and a moderate probability that the relationship is causal
2-
Case-control or cohort studies with a high risk of confounding, bias, or chance and a significant
risk that the relationship is not causal
3 Non-analytic studies, e.g., case reports, case series
4 Expert opinion
Scottish Intercollegiate Guidelines Network (SIGN)
Revised grading system for recommendations in
evidence-based guidelines

Level Evidence
1++ High-quality meta-analyses, systematic reviews of RCTs, or RCTs with a very low risk of bias
1+ Well-conducted meta-analyses, systematic reviews of RCTs, or RCTs with a low risk of bias
1- Meta-analyses, systematic reviews or RCTs, or RCTs with a high risk of bias
2++
High-quality systematic reviews of case-control or cohort studies or high-quality case-control or
cohort studies with a very low risk of confounding, bias, or chance and a high probability that
the relationship is causal
2+
Well-conducted case-control or cohort studies with a low risk of confounding, bias, or chance
and a moderate probability that the relationship is causal
2-
Case-control or cohort studies with a high risk of confounding, bias, or chance and a significant
risk that the relationship is not causal
3 Non-analytic studies, e.g., case reports, case series
4 Expert opinion
Scottish Intercollegiate Guidelines Network (SIGN)
Revised grading system for recommendations in
evidence-based guidelines

Features of RCT’s

Features of RCT’s
 Rarticipants are randomly assigned

Features of RCT’s
 Rarticipants are randomly assigned
 Comparative - Treatment group compared with
control group with similar characteristics

Features of RCT’s
 Rarticipants are randomly assigned
 Comparative - Treatment group compared with
control group with similar characteristics
 Allocation concealment

Features of RCT’s
 Rarticipants are randomly assigned
 Comparative - Treatment group compared with
control group with similar characteristics
 Allocation concealment
 Blinding

Features of RCT’s
 Rarticipants are randomly assigned
 Comparative - Treatment group compared with
control group with similar characteristics
 Allocation concealment
 Blinding
 Is the most scientifically rigorous method of
hypothesis testing  gold standard

Bias in RCT’s
Type of bias Prevention / elimination of bias
Selection bias Proper randomization
Performance bias
Blinding of participants and people
administering treatment
Detection bias Blinding of outcome assessors/analyzers
Attrition bias Intention-to-treat analysis
Publication bias Trial registration, no selective reporting

Bias in RCT’s
Type of bias Prevention / elimination of bias
Selection bias Proper randomization
Performance bias
Blinding of participants and people
administering treatment
Detection bias Blinding of outcome assessors/analyzers
Attrition bias Intention-to-treat analysis
Publication bias Trial registration, no selective reporting

Bias in RCT’s
Type of bias Prevention / elimination of bias
Selection bias Proper randomization
Performance bias
Blinding of participants and people
administering treatment
Detection bias Blinding of outcome assessors/analyzers
Attrition bias Intention-to-treat analysis
Publication bias Trial registration, no selective reporting

Bias in RCT’s
Type of bias Prevention / elimination of bias
Selection bias Proper randomization
Performance bias
Blinding of participants and people
administering treatment
Detection bias Blinding of outcome assessors/analyzers
Attrition bias Intention-to-treat analysis
Publication bias Trial registration, no selective reporting

Bias in RCT’s
Type of bias Prevention / elimination of bias
Selection bias Proper randomization
Performance bias
Blinding of participants and people
administering treatment
Detection bias Blinding of outcome assessors/analyzers
Attrition bias Intention-to-treat analysis
Publication bias Trial registration, no selective reporting

Bias in RCT’s
Type of bias Prevention / elimination of bias
Selection bias Proper randomization
Performance bias
Blinding of participants and people
administering treatment
Detection bias Blinding of outcome assessors/analyzers
Attrition bias Intention-to-treat analysis
Publication bias Trial registration, no selective reporting

Systematic Review
 A research article that,

Systematic Review
 A research article that,
o Identify relevant studies

Systematic Review
 A research article that,
o Identify relevant studies
o Appraises their quality

Systematic Review
 A research article that,
o Identify relevant studies
o Appraises their quality
o Summarizes their results
using scientific methodology

Meta-analysis
Systematic
Review Systematic ReviewMeta-Analysis

Meta-analysis
 The use of statistical techniques to combine the results of studies
addressing the same question into a summary measure
Systematic
Review Systematic ReviewMeta-Analysis

Meta-analysis
 The use of statistical techniques to combine the results of studies
addressing the same question into a summary measure
 Increases the size of the `overall sample’  enhances statistical power
Systematic
Review Systematic ReviewMeta-Analysis

Meta-analysis
 The use of statistical techniques to combine the results of studies
addressing the same question into a summary measure
 Increases the size of the `overall sample’  enhances statistical power
 Quantitative synthesis  only when the studies to be combined are
clinically and statistically homogeneous
Systematic
Review Systematic ReviewMeta-Analysis

Meta-analysis
 The use of statistical techniques to combine the results of studies
addressing the same question into a summary measure
 Increases the size of the `overall sample’  enhances statistical power
 Quantitative synthesis  only when the studies to be combined are
clinically and statistically homogeneous
 Systematic Review ≠ Meta-Analysis
Systematic
Review Systematic ReviewMeta-Analysis

What are Cochrane reviews?

What are Cochrane reviews?
 The Cochrane Collaboration is an international
organization that aims to organize medical research
information in a systematic way promoting the
accessibility of systematic reviews of the effects of health-
care interventions

What are Cochrane reviews?
 The Cochrane Collaboration is an international
organization that aims to organize medical research
information in a systematic way promoting the
accessibility of systematic reviews of the effects of health-
care interventions
 Systematic reviews of primary research in human health
care and health policy, undertaken by members of The
Cochrane Collaboration adhering to a specific
methodology

Preferred Reporting Items for Systematic
Reviews and Meta-Analyses (PRISMA)

Assessing the quality
Assessment of Multiple Systematic Reviews - AMSTAR

This article is the result of a debate at the European Journal of
Orthodontics Open Session in 2013 in Reykjavik, Iceland

Part I: RCTs are for orthodontics
Lars Bondemark

Part I: RCTs are for orthodontics
 The first published RCT 1948 ‘Streptomycin treatment
of pulmonary tuberculosis’
Lars Bondemark

Part I: RCTs are for orthodontics
 The first published RCT 1948 ‘Streptomycin treatment
of pulmonary tuberculosis’
Orthodontic RCTs per year between 1990 and 2013
Lars Bondemark

Randomization

Randomization
 Ensures both known and unknown
determinants are evenly distributed
among the different study groups

Randomization
 Ensures both known and unknown
determinants are evenly distributed
among the different study groups
 Minimizes bias in assessment of
differences in effects between two
or more treatment alternatives
Confounding factors equally
distributed

Randomization
 Ensures both known and unknown
determinants are evenly distributed
among the different study groups
 Minimizes bias in assessment of
differences in effects between two
or more treatment alternatives
Confounding factors equally
distributed
 Idea with randomization is that the
treatment will be the only thing that
will constitute a significant
difference between the patient
groups

Randomization
 Ensures both known and unknown
determinants are evenly distributed
among the different study groups
 Minimizes bias in assessment of
differences in effects between two
or more treatment alternatives
Confounding factors equally
distributed
 Idea with randomization is that the
treatment will be the only thing that
will constitute a significant
difference between the patient
groups
 Best if done by an independent
person

CONSORT
 Consolidated Standards of Reporting Trials

Recommendations

Recommendations
 Create a relevant question use PICO

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome
e.g. Is Quad-helix treatment (intervention) more cost-effective (outcome) than platal
expansion (control) in 8–10 year-old patients with unilateral posterior crossbite
(population)?

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome
e.g. Is Quad-helix treatment (intervention) more cost-effective (outcome) than platal
expansion (control) in 8–10 year-old patients with unilateral posterior crossbite
(population)?
 Use proper randomization

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome
e.g. Is Quad-helix treatment (intervention) more cost-effective (outcome) than platal
expansion (control) in 8–10 year-old patients with unilateral posterior crossbite
(population)?
 Use proper randomization
 Plan and coordinate the trial carefully - have patience since it takes time to run an RCT

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome
e.g. Is Quad-helix treatment (intervention) more cost-effective (outcome) than platal
expansion (control) in 8–10 year-old patients with unilateral posterior crossbite
(population)?
 Use proper randomization
 Plan and coordinate the trial carefully - have patience since it takes time to run an RCT
 Have sufficient sample size – if small, greater risk of unknown or confounding factors that
may interfere the outcomes

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome
e.g. Is Quad-helix treatment (intervention) more cost-effective (outcome) than platal
expansion (control) in 8–10 year-old patients with unilateral posterior crossbite
(population)?
 Use proper randomization
 Plan and coordinate the trial carefully - have patience since it takes time to run an RCT
 Have sufficient sample size – if small, greater risk of unknown or confounding factors that
may interfere the outcomes
 Use valid and reliable methods

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome
e.g. Is Quad-helix treatment (intervention) more cost-effective (outcome) than platal
expansion (control) in 8–10 year-old patients with unilateral posterior crossbite
(population)?
 Use proper randomization
 Plan and coordinate the trial carefully - have patience since it takes time to run an RCT
 Have sufficient sample size – if small, greater risk of unknown or confounding factors that
may interfere the outcomes
 Use valid and reliable methods
 Follow the CONSORT statement

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome
e.g. Is Quad-helix treatment (intervention) more cost-effective (outcome) than platal
expansion (control) in 8–10 year-old patients with unilateral posterior crossbite
(population)?
 Use proper randomization
 Plan and coordinate the trial carefully - have patience since it takes time to run an RCT
 Have sufficient sample size – if small, greater risk of unknown or confounding factors that
may interfere the outcomes
 Use valid and reliable methods
 Follow the CONSORT statement
 Use blinding if possible

Recommendations
 Create a relevant question use PICO
• P - population
• I - intervention
• C - control
• O - outcome
e.g. Is Quad-helix treatment (intervention) more cost-effective (outcome) than platal
expansion (control) in 8–10 year-old patients with unilateral posterior crossbite
(population)?
 Use proper randomization
 Plan and coordinate the trial carefully - have patience since it takes time to run an RCT
 Have sufficient sample size – if small, greater risk of unknown or confounding factors that
may interfere the outcomes
 Use valid and reliable methods
 Follow the CONSORT statement
 Use blinding if possible
 Use the Intention-to-treat (ITT)analysis to evaluate the results

Intention-to-treat Analysis
https://www.youtube.com/watch?v=nnxg0FJwPjY

Part II: RCTs are not for orthodontics
Sabine Ruf

Part II: RCTs are not for orthodontics
 New EBM ideology - RCTs may minimize, but do not eliminate
bias
Sabine Ruf

Why RCTs are not ‘working’ for crucial
clinical orthodontic questions?

Why RCTs are not ‘working’ for crucial
clinical orthodontic questions?
 Primary goal of RCT  test whether an intervention works
by comparing it to a control condition

Why RCTs are not ‘working’ for crucial
clinical orthodontic questions?
 Primary goal of RCT  test whether an intervention works
by comparing it to a control condition
o No treatment
o A placebo treatment
o An alternative treatment

Why RCTs are not ‘working’ for crucial
clinical orthodontic questions?
 Primary goal of RCT  test whether an intervention works
by comparing it to a control condition
o No treatment
o A placebo treatment
o An alternative treatment
 Orthodontics is a device-driven specialty  not a series of
pills that can be administered at random and evaluated
blindly

Why RCTs are not ‘working’ for crucial
clinical orthodontic questions?
 Primary goal of RCT  test whether an intervention works
by comparing it to a control condition
o No treatment
o A placebo treatment
o An alternative treatment
 Orthodontics is a device-driven specialty  not a series of
pills that can be administered at random and evaluated
blindly
 Even ‘invisible’ appliances are visible to the pt & operator

Why RCTs are not ‘working’ for crucial
clinical orthodontic questions?
 Primary goal of RCT  test whether an intervention works
by comparing it to a control condition
o No treatment
o A placebo treatment
o An alternative treatment
 Orthodontics is a device-driven specialty  not a series of
pills that can be administered at random and evaluated
blindly
 Even ‘invisible’ appliances are visible to the pt & operator
 Excludes the possibility of orthodontic placebo treatments
at least for the majority of the questions

Effect of Informed Consent

Effect of Informed Consent
 RCT by Bergmann et al. in 1994,
shortly before informed consent
became mandatory in France

Effect of Informed Consent
 RCT by Bergmann et al. in 1994,
shortly before informed consent
became mandatory in France
 Aim: Effect of informed consent on
analgesic activity of pain killers

Effect of Informed Consent
 RCT by Bergmann et al. in 1994,
shortly before informed consent
became mandatory in France
 Aim: Effect of informed consent on
analgesic activity of pain killers
 49 consecutively hospitalized patients
with mild to moderate cancer pain

Effect of Informed Consent
 RCT by Bergmann et al. in 1994,
shortly before informed consent
became mandatory in France
 Aim: Effect of informed consent on
analgesic activity of pain killers
 49 consecutively hospitalized patients
with mild to moderate cancer pain
 Uninformed group - 25 received both
treatments without any information

Effect of Informed Consent
 RCT by Bergmann et al. in 1994,
shortly before informed consent
became mandatory in France
 Aim: Effect of informed consent on
analgesic activity of pain killers
 49 consecutively hospitalized patients
with mild to moderate cancer pain
 Uninformed group - 25 received both
treatments without any information
 Informed-consent group - 24 had a
complete information about the trial;
6 refused to participate 18 pt’s

Effect of Informed Consent
 RCT by Bergmann et al. in 1994,
shortly before informed consent
became mandatory in France
 Aim: Effect of informed consent on
analgesic activity of pain killers
 49 consecutively hospitalized patients
with mild to moderate cancer pain
 Uninformed group - 25 received both
treatments without any information
 Informed-consent group - 24 had a
complete information about the trial;
6 refused to participate 18 pt’s
 VAS score of pain before and 30, 60,
120 and 180 min after the intake of
naproxen and placebo were
recorded

Changes (mm) in VAS pain levels after naproxen or placebo intake in patients
with and without informed consent

Changes (mm) in VAS pain levels after naproxen or placebo intake in patients
with and without informed consent
Pain reduces with the drug in the
non-informed group
Pain increases after receiving the
placebo in the non-informed group

Changes (mm) in VAS pain levels after naproxen or placebo intake in patients
with and without informed consent
Pain reduces in both the
drug and placebo in the
informed consent group

 First described in the 1950s by Henry A. Landsberger after experiments conducted at the
Hawthorne works electric company to determine if increasing or decreasing the amount of
light workers received would have an effect on worker productivity
 Employee productivity seemed to increase due to the changes, but then decreased at
after the experiment was over
 Researchers suggested that productivity increased due to attention from the research team
and not because of changes to the experimental variables
 Lansdberger defined the Hawthorne effect as a short-term improvement in performance
caused by observing workers

All orthodontic treatment modalities requiring cooperation cannot be
tested reliably using a RCT design, at least not if we are seeking clinically
relevant truth. And that is maybe ‘why RCTs in orthodontics have not
achieved their intended objective

Drawbacks of RCT’s
 High costs
 Ethical problems
 Informed consent
 Bias problems
 Clinicians/patients preference for certain treatment
 Recruiting sufficient patients

Final Thoughts
 Lack of evidence does not necessarily imply lack of effect
 Instead, there is need for further relevant evaluations. Clearly,
new well-designed RCTs and non-randomized studies can
achieve important support to reliable evidence in orthodontics