1. Principles of Urologic Research
& Evidence Based Urology
A Presentation
by
Farouk Ashraf Hammoud
2. First and foremost I would like to thank Allah
for giving me the power to accomplish this
work.
I would also like to express my gratitude to my
supervisors, Prof. Dr. Mohammed Shoeb,
Dr. Tarek Osman, and Dr. Diaa Abdelftah,
for their continuous supervision and
irreplaceable guidance. Also Prof. Dr. Amr El
Shorbagy and Prof. Dr. Khaled El Gamal,
thank you for attending, I know I put you
through some difficulties.
Finally my deep thanks goes out to my family
and friends for their constant support.
4. Ancient medicine:
Based on superstition
and medical theories
Renaissance period:
(14th to 17th century)
Revolutionary scientific
changes
5. Medicine & Statistics Change Research
Bradford Hill Archie Cochrane
First modern randomized clinical Founding father of EBM
6. What is EBM?
“The conscientious, explicit and judicious use of
current best evidence in making decisions about
the care of individual patients.”
The integration of individual clinical expertise with the best
available external clinical evidence from systematic
research.
8. Conducting Research
The EBM process can be summarized by
four steps:
(1) Formulate a sensible, focused clinical
question: PICOT
(2) Search the medical literature for evidence
related to the focused, clinical question.
(3) Rate the quality of the available studies.
(4) Apply the evidence to a particular patient
or clinical situation.
9. PICOT
P = Population.
I = Intervention.
C = Comparison.
O = Outcome.
T = Type of study. Choose the best study
design
10. PICOT
P = 63 year old with BPH
I = medical therapy
C = surgery
O = reducing lower urinary tract symptoms
T = in a randomized controlled trial or
systematic review of randomized controlled
trials
12. Where to Search
Synthesized sources: textbooks, guidelines
Selected sources: research that is filtered
and published as database Ex. BMJ EBM
Bibliographic database of published journal
articles. Ex. PubMed
13. Type Name URL Info
Database Cochrane Library http://www.thecochranelibrary.com Database of systemic
reviews
Medline/Pubmed http://www.pubmed.gov Major source of
biomedical publications
EMBASE http://www.embase.com/ European equivalent of
Medline
TRIP: Turning Research Into http://www.tripdatabase.com/ Searches E-Medicine,
Practice guidelines, & other
evidence-based
publications
Guidelines American Urological http://www.auanet.org/guidelines/ Practice guidelines for
Association Guidelines patient management
and current standard
European Association of http://www.uroweb.org/
of care
Urology/guidelines
Preappraised Evidence-base Medicine http://ebm.bmj.com/ Provides
sources ACPJC: American College of http://www.acpjc.org structured abstract of
articles selected from
Physicians Journal Club
journals by an editorial
board, plus critical
appraisal of the validity,
importance, and
applicability
CENTRAL: Cochrane Central http://www.mrw.interscience.wiley.com/coch Non English studies
Register of Controlled Trials rane/cochrane_clcentral_articles_fs.html included
15. Types of Studies
Descriptive Studies
Observational Analytic Studies
◦ Cross Sectional studies
◦ Case Control studies
◦ Cohort studies
Experimental Studies
◦ Randomized controlled trials
16. Hierarchy of Study Types
Descriptive Analytic
•Case report
•Case series
•Survey Observational Experimental
•Cross sectional •Randomized
•Case-control controlled trials
•Cohort studies
Strength of evidence for causality between a risk factor and outcome
18. Descriptive Studies
Surveys
“How many men in the U.S. filled Viagra
prescriptions in 2004?”
Case reports or case series
Sildanefil-associated serous macular detachment.
19. Descriptive Studies
Cannot establish causal relationships
Play an important role in generating hypotheses
to be tested
Alerting physicians to a new disease
◦ Squamous cell carcinoma in sexual partner of Kaposi sarcoma
patient. Lancet. 1982 Jan 30;1(8266):286.
◦ New outbreak of oral tumors, malignancies and infectious
diseases strikes young male homosexuals. CDA J. 1982
Mar;10(3):39-42.
◦ AIDS in the "gay" areas of San Francisco. Lancet.
1983 Apr 23;1(8330):923-4.
20. Hierarchy of Study Types
Descriptive Analytic
•Case report
•Case series
•Survey Observational Experimental
•Cross sectional •Randomized
•Case-control controlled trials
•Cohort studies
Strength of evidence for causality between a risk factor and outcome
25. Cross Sectional Study:
Descriptive value:
◦ How many medical students smoke cigarettes?
◦ What is the age and sex distribution of medical students who
smoke cigarettes?
Analytic value:
◦ Is there an association between regular cigarette smoking and
test scores among med students?
28. Cohort Study: Strength
Exposure variable was present before
outcome variable
Incidence
Multiple outcomes of a single exposure
Retrospective cohorts where available are
cheaper and quicker.
Analyse predictors (risk factors) thereby
enabling calculation of relative risk.
29. Measuring Relative Risk
Risk ratio
Disease (relative risk)
Yes No A
A+B
C
C+D
Yes A B
Risk
Factor
No C D
30. Cohort Study: Weakness
Confounding variables
Selection bias
Loss to follow up
Recall bias
33. Case-Control Study: Structure
TARGET CASES
ACTUAL CASES
Males with prostate cancer
Males with prostate cancer confirmed
Dietary Zinc intake by biopsy
YES
Dietary Zinc intake
NO ACTUAL CONTROLS
TARGET admitted to hospital for non-
Males CONTROLS
Males withoutand non-hormone related
neoplastic prostate cancer
disease
time
34. Case Control Study: Strength
• Simple, fast and inexpensive
• Rare outcome & long latency period
• Determines the strength of the
association between each predictor
variable and the presence or absence
of disease
• Determines odds ratio
35. Case Control Study: Weakness
Can only look at one outcome
Confounding variables
Bias is a major problem
◦ Selection
◦ Recall
36. Hierarchy of Study Types
Descriptive Analytic
•Case report
•Case series
•Survey Observational Experimental
•Cross sectional •Randomized
•Case-control controlled trials
•Cohort studies
Strength of evidence for causality between a risk factor and outcome
38. RCTs:
Gold standard: Causality
Investigator controls the predictor variable
(intervention or treatment)
Randomization:
◦ The only way to avoid selection and
confounding bias
40. Guarding against Bias: Analysis
Sample size
Reporting effect size
Intention to treat principle
41. Sample Size
Optimal sample size
◦ Type I Error: False positives < 5%
P-value < 0.05
Probability that H0 is rejected
◦ Type II Error: False negatives > 20%
Less than 1/20 = statistically significant
42. P-value & CI
Confidence Interval relates to P-value
◦ Confidence level = 95%
◦ Confidence limits = Range
Wider CI reflects higher confidence level
◦ Ex. 99% CI when mean age CI = 25 and 29
43. Effect Size
The observed relationship between an
intervention and an outcome is
statistically expressed as an “estimate of
effect”
◦ Odds ratio (OR) or a
◦ Relative risk (RR)
True effect size lies within confidence
interval
44. Intention-to-treat analysis:
Including and analysing all randomised
patients according to their original
treatment allocation, irrespective of
whether they actually received or completed
the treatment.
This preserves the unbiased comparison of
treatment groups afforded by
randomization.
47. Systematic Review
Attempts to review all of the relevant
articles in the field
Uses study design based on the hierarchy of
evidence
◦ Most use RCTs (gold standard)
Uses clearly defined parameters
48. Systematic Review
Accurate and reliable
Reduces large quantities of information
into manageable size
External validity is more applicable than
populations from single studies
50. Meta-analysis
Type of systematic review that focuses on
numerical results
◦ Uses statistics to integrate findings
Can detect small effects across a group of
studies that individually lack statistical
power to detect the effect
Common problem is heterogeneity of
studies
◦ Noncompatibility
54. EBM
Level Therapy/Prognosis/Aetiology
1a SR (with homogeneity) of RCTs
1b Individual RCT (with narrow Confidence Interval)
1c All or none case report/series
2a SR (with homogeneity) of cohort studies
2b Cohort study
3a SR (with homogeneity) of case-control studies
3b Case-Control Study
4 Case-series
5 Expert opinion without explicit critical appraisal, or based on
physiology, bench research or "first principles"
GRADE system: Grading of Recommendations Assessment, Development & Evaluation
55. EBM - What is it?
Clinical
Expertise
Research Patient
Evidence Preferences
56. EBM Method
Ask clinical
questions
Acquire the Appraise
best evidence the evidence
Apply
evidence to
patient care
57. Ask Clinical Questions
P I C O
Patient/ Intervention/
Population Comparison Outcome
Exposure
does
In patients with compared to decrease LUTS?
treatment
BPH surgery
with alpha
blocker
58. Acquire the Best Evidence
Where do you find high-quality evidence?
◦ Textbook (print or online)
◦ Pre-appraised evidence
Best Evidence
Clinical Evidence (Therapy only)
Cochrane Collaboration (Therapy only)
UpToDate
◦ Medline or PubMed search: find and review
articles
59. Appraise the Evidence
Are the results valid?
What are the results?
Can we apply the results to our patient?
60. Appraise the Evidence
Determine if evidence is unbiased
or flawed
◦ Critically appraise articles yourself
◦ Used a source that appraises trials for you
Best Evidence
Clinical Evidence
Cochrane Library
UpToDate
61. Apply the Evidence
Evidence must be applied to each
individual patient
◦ Is your patient similar enough to those
studied?
◦ Do benefits outweigh harms?
◦ Cost
◦ What are your patient’s values and
preferences?
62. Take Home Message
Research is like riding a bicycle
◦ It needs practice
Research needs planning
◦ If you fail to plan, you plan to fail
Not all of us need to do research
◦ But all of us should practice EBM
P:The patient’s characteristics, including age, gender, and condition, plus other relevant clinical or medical history featuresI: What intervention are you considering evaluating or using? Describe the main intervention, exposure, test, or prognostic factor under consideration. In the case of treatment, this might be the standard of care or a placebo, or a “gold standard” diagnostic test. A Comparison is not always needed. Describe what is hoped to be achieved, measured or affected
This is ur example.Say wat picot stands for.
After formulating the question.. The search begins.
Begins with: synthesized (campbell, aua/eau guidelines, e-medicine) + cochrane library – db of SR.. Under urology: subcateg (cancer, stones, incontinence) SR of all relevant articles pertaining to topic so back to ex. Urology: obstructive: prostate: different reviews for finasteride, tamsulosin, terazosin ttt for bphThen search databases ie BMJ EBM for relevant articles (case reports, cohorts, etc) which are peer-reviewed by a panel of clinical experts who rate the importance of each studyIf unsatisfactory results, fallback is larger database of journal articles (pubmed – over 5000 internation med journals)
Descriptive studies give us a picture of what is happening in a population.Analytic studies attempt to establish a link btwn an exposure and an outcome.
Descriptive shows incidence (# of new cases in time interval), prevalence (# of cases at point in time OR % as cases/population)Analytical observational: researcher measures the outcome Experimental: researcher manipulates outcome<> Evidence is stronger when studies can show relation btwn cause & effect ( e as consequence of c)
Surveys give a general outlook on a population ie questionnaire, Case reports or series (individual and collection of similar reports)
Types of analytic observationalEach has an objective as we will see later
(in direction of time: cross sec – same point in time, case cont – retrospective, cohort – prospective)
Snapshot in time. Exposure & outcome measured simultaneously. Impossible to distinguish which came first.
Descriptive value shows prevalence. Also can establish association, but not cause & effect.This is usually confirmed by longitudinal studies..
Cohort is example of longitudinal study. This study Follows subjx over time to see who will develp outcome and who will not (incidence)<>(prospective or retro)Looksfwd in time exposure to outcome.. Or when data is already collected or exposure would be unethical (smoking) is done retro.
Since cohort studies measure potential causes before the outcome has occurred.. Exposure precedes outcome. Cause is somewhat established. CFVIncidence is new cases that develop over time interval.RR is probability of developing outcome if exposed vs if not exposed
Risk factor = exposureDisease = outcome<> To calc RR..RR =1 no diff RR <1 = exposed group less likely to develop disease RR >1 = exposed group more likely to develop
CFV = all other factors that differ btwn groups (exp and nonexp) Ex. Asthma (exp) lung cancer (outcome)…. Protects? Don’t think so. Asthmatics don’t smoke. Smoking is CFV.Selection = sample is not representative of the pop. (employed ppl better health than unemployed)Losing contactRecall = in retro, ppl more likely to remember certain events they believe to be risk factors
Works backwards. Starts with outcome. Compares to control group w/o outcome. Then determines retro whether exposure prevalence is higher in cases than controls. So take for example pts w prostate cancer .. And lets say for example dietary factors in carcinogenesis.
Simple = questionnaireIn rare cases, ie new variant CJD, limited cases present; and long latency time to develop disease; preferrable to work retrospectiveCase control studies gives researcher clues as to which factors are assoc w disease (predictor variable) OR = probability of even happening vs not happening (odds of exposed group developing outcome vs odds of non exposed gp developing outcome)1 = no diff. btwnppl in group w event and group w/o event
Gold stand. = in establishing causal effects in studies of therapyRandomly assigning subjx to groups prevents selection biasPrevents CFV by a randomised controlled trial (RCT)—the investigator experiments with the effect of the exposure by assigning exposure to a random sample of the study subjects.the randomisation process allows the investigator to assume that not only known, but also unknown, potential confounders are distributed evenly among the exposed and the unexposed. Therefore, they are not associated with the exposure, hence they cannot be confounders.
Several ways rcts guard against biasRndm: to balance btwn known & unknown factorsSimple = coin toss Urn= drawing balls Blinding:pts or researcher: pts AND researcher: subjct, researcher, person treating
One of most important safeguards is choosing sample size necessary to detect clinically significant data.
Optimal size is calc by comp. balance btwn type I and II errors. Sample size shud be so that Max prob of making type 1; 5% or less. Max false negative; 20% or lessImp bc more subjx r needed to lower error rate of type II. Inadequate sample size can show no diff. when in fact there is.<>Relates to (Pvalue) Prob H0 is rejected. 1/20 or less = statistically significant = 1/20 that difference is dt chance thus rejecting null hypothesis H0 (no diff)
95% level corresponds to hypothesis testing with p-values less than 0.05Means that 95% of time we expect true value to be between upper & lower confidence limitsEx. If the value is mean age of first birth in women w bleeding disorders; CI = 25.5 and 28.4; there4 we are 95% confident that the true mean age lies btwn 25.5 and 28.4
Large trials can cost millions of dollars. Can take a long period to conduct. Also exposing pts to inferior ttt is thought to be unethical.Can not be generalized to broader community.. Usually dt strict selection criteria.
Reviews all study designs, but focuses mostly on RCTs as they are gold standard, and give most unbiased outcome
Quality…. And quantity.
Uses hierarchy of evidence and grades them in levelsThis scheme was developed at the centre for EBM at oxford university and uses the GRADE system.For studies of therapy/prognosis/aetiology.. This is brief description.