Your SlideShare is downloading. ×
Medical Literature
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×

Introducing the official SlideShare app

Stunning, full-screen experience for iPhone and Android

Text the download link to your phone

Standard text messaging rates apply

Medical Literature

693
views

Published on


0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
693
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
26
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. Medical Literature: Reading, Interpreting, and Writing Muhammad Mamdani, PharmD, MA, MPH Director, Applied Health Research Centre LKSKI of St. Michael’s Hospital Associate Professor – University of Toronto Adjunct Faculty – King Saud University February 2010
  • 2. Objectives
    • To better understand the basic purpose of research and the context under which research is conducted
    • To review basic clinical research study designs
    • To review two standard study review tools
    • To review basic principles in writing clinical research manuscripts
  • 3. Clinical Research: Basic Purpose and Context
  • 4. The Common Goal of Research
    • ‘ The’ Goal
      • Attempt to estimate the ‘truth’
    TRUTH Study 3 Study 1 Study 2
  • 5. Perception: Reality vs Belief
  • 6. Variability in Perceptions
  • 7. The Trouble with Research Example: H. pylori eradication and non-ulcer dyspepsia
    • McColl et al (NEJM, 1998)
      • Treatment (n=154): Omeprazole 20 mg bid + amoxicillin 500 mg tid + 400 mg metronidazole tid x 2 weeks
      • Comparison (n=154): omeprazole 20 mg bid x 2 weeks
      • Symptom resolution at 1 year: Tx = 21% vs. Comparison = 7% (p<0.001)
    • Blum et al (NEJM, 1998)
      • Treatment (n=164): Omeprazole 20 mg bid + amoxicillin 1000 mg bid + 500 mg clarithromycin bid x 1 week
      • Comparison (n=164): omeprazole 20 mg bid x 1 week
      • Symptom resolution at 1 year: Tx = 27% vs. Comparison = 21% (p=0.17)
  • 8. The Evolving Nature of Research
  • 9. The Evolving Nature of Research
    • Analyzed 115 articles published in 1990-2003 in the 3 major general medical journals (NEJM, JAMA, Lancet) and specialty journals that had received over 1000 citations each by August 2004
    • 49 reported evaluations of health care interventions; 45 claimed that the interventions were effective.
    • By 2004 5/6 non randomised studies and 9/39 randomised trials were already contradicted or found to be exaggerated
      • Ioannidis JP. JAMA 2005
  • 10. But How Do I Know About the Quality and Validity of the Study?
  • 11. Understanding Clinical Research: What do I Need to Fully Interpret a Clinical Study?
    • Understand Clinical Research Methodology: Epidemiology
    • Analytical Expertise
      • Quantitative: Biostatistics
      • Qualitative: Qualitative Analytics
    • Clinical Sense
      • Medical / Healthcare Training
  • 12. The Medical Research Spectrum (Sung, JAMA, 2003)
  • 13. The Medical Research Spectrum (Sung, JAMA, 2003)
  • 14. Basic Clinical Research Methodology
  • 15. Comparative Clinical Research: A Simplistic Overview (Melot, CCM, 2009)
  • 16. Approaches to Clinical Research
    • Two Major Approaches
      • Experimental Studies: Randomized Trials
        • Randomly allocate subjects to different treatment strategies and follow them up to assess outcomes
        • Operationally complex, analytically simple
      • Observational Studies
        • Operationally simple, analytically complex
  • 17. Randomized Clinical Trial: Design (Melot, CCM, 2009)
  • 18. Validity and Biases in Clinical Trials (Melot, CCM, 2009)
  • 19. Major Observational Study Types
    • Cross-sectional Study
    • Cohort Study
    • Case-Control Study
    • Case-Cross-Over Study
  • 20. TIME Basic Schematic for Cross-Sectional Study Data for analysis
    • Major Issues:
    • Temporal Sequencing
    • Selection Bias
    • Confounding
  • 21. TIME Look-back Window Index Entry Date for Exposure End of Follow-up Date Observation Window Basic Schematic for Cohort Study
    • Major Issues:
    • Selection Bias
    • Confounding
  • 22. Basic Schematic for Case-Control Study Time Period Event No Event Cases Controls Time Period
    • Major Issues:
    • Selection Bias
    • Confounding
    • Only indirect estimates of time effects
  • 23. Basic Schematic for Case-Crossover Study Time Period A Time Period B Event Compare exposure in Time Period A vs. Time Period B only among patients with an event and exposure in either period
    • Major Issues:
    • Largely used for exposures with immediate effects
    • Small number
    • ‘ Reverse protopathic’ bias
  • 24. How Do I Know Which Study Design is Best? Oxford Centre for Evidence-Based Medicine, 2002 Level of Evidence Study Type Level 1 RCTs Level 2 Cohort Studies Level 3 Case-Control Studies Level 4 Case Series Level 5 Expert Opinion
  • 25. Analytical Aspects
  • 26. Numeric Literacy (Horton and Switzer, NEJM, 2005)
    • Sophistication of statistical methods or articles published in the NEJM has been increasing over time
    • Some statistical tests
      • T-tests
      • Contingency tables
      • Non-parametric tests
      • Epidemiologic statistics
      • Pearson’s correlation
      • Nonparametric correlation (e.g. spearman’s correlation)
      • Simple linear regreasion
      • Analysis of variance
      • Transformations
    • Only 21% of articles published between January 2004 – June 2005 would be considered accessible by those with basic training in biostatistics
  • 27. Numerical Methods
    • Risks: Proportions of Patients
      • Absolute Risk (AR)
      • Number Needed to Treat (NNT)
    • Ratio of Risks
      • Relative Risk (RR)
      • Hazard Ratio (HR)
      • Odds Ratio (OR)
  • 28. Measures of Association: An Example Total n = 40,000 N per study group = 20,000 Absolute Risks AR (Placebo) = 0.5% AR (Treatment) = 0.15% ARR = 0. 5% - 0.15% = 0.35% NNT = 1/0.35% = 296 Relative Risks RR = 0.15% / 0.5% = 0.30 RRR = 1/RR = 70% OR (CC: Logistic Regression) HR (Cohort: Cox PH Models) Placebo Treatment MI 100 30 No MI 19900 19970
  • 29. Issues with Numerical Approaches
    • Lacy et al, Am J Cardiol, 2001
      • Assessed 400 health professionals for their willingness to prescribe a drug based on different measures for reporting the same likelihoods
  • 30. Literature Evaluation Tools
  • 31. Standards for Assessing Quality of Research
    • Clinical Trials
      • CONSORT statement
      • http://www.consort-statement.org/
    • Observational Research
      • STROBE Statement
      • http://www.strobe-statement.org/
  • 32. Consort Statement
  • 33. STROBE
  • 34. Group Activity: Pitt et al
  • 35. Group Exercise
    • 20 Minutes Group Review
    • One person from each group to present for up to 2 minutes
    • Use CONSORT
    • Groups
      • Group 1: Abstract and Introduction
      • Group 2: Methods
      • Group 3: Results
      • Group 4: Comment – make recommendation
  • 36. Pitt et al: Study Overview
  • 37. RALES: Randomized ALdactonE Study Pitt et al, N Engl J Med, 1999
    • Double-blind randomized controlled trial in patients with congestive heart failure
    • Primary endpoint: all-cause mortality
    Screened (n=?) Randomize Spironolactone (n=822) 25 mg po qd - Increase to 50 mg po qd where tolerated Placebo (n=841) Inclusion Exclusion
  • 38. RALES Criteria
    • Inclusion / Exclusion
      • NYHA class III or IV at time of enrolment
      • LVEF < 35% within 6 months
      • Exclude patients with serum creatinine  2.5 mg/dL or serum potassium > 5 mmol/L
    • Follow-up
      • Lab and clinic follow-up at 4 weeks and 3 and 6 months
      • Appropriate use of ACE inhibitors and beta-blockers
      • D/c K-sparing diuretics and K + supplements
      • Holding spironolactone for hyperkalemia or creatinine > 4 mg/dL
  • 39. RALES - Results Outcome Absolute Risk Relative Risk Death Pl=46% Spir=35% p<0.001 0.70 (0.60-0.82) Readmission for HF Pl=36% Spir=26% p<0.001 0.65 (0.54-0.77) Serious Hyperkalemia Pl=1.2% Spir=1.7% p=0.42 NS
  • 40. What Happened in Actual Practice? Juurlink et al, NEJM, 2004
  • 41. Spironolactone Prescription Uptake 1994-2001 RALES
  • 42. Hospital Admission Associated with  K + RALES
  • 43. Admission Associated with  K + Ending in Death RALES
  • 44. What about the expected benefits?
  • 45. Readmission for Heart Failure RALES
  • 46. All-Cause Mortality RALES
  • 47. What Happened???
  • 48. In the ‘Real World’….
    • Different patients may get the drug
      • > 50% of population that would normally use the drugs in clinical practice often do not meet inclusion / exclusion criteria of RCTs (e.g. Gill et al, CJCP, 2004)
    • Patients may not be monitored as carefully
    • Patients may not take their drug as they should
    • Patients may take interacting drugs that they shouldn’t be on
    • Patients may not adhere to their drugs optimally
  • 49. What’s the Problem?
  • 50. Anton et al J Clin Pharm Ther 2003; 28: 285-7
    • Anton et al. J Clin Pharm Ther 2003;28:285-7
    • Retrospective cohort
      • Single hospital
      • 110 patients Rx spironolactone + ACEI
        • Mean age 71 years, half with DM
      • Outcomes
        • Cessation of spironolactone
        • Hyperkalemia
  • 51. Findings
    • 24% developed  K + (> 5.5 mEq/L) at 1 year
    • Many stopped spironolactone
    J Clin Pharm Ther 2003;28:285-7
  • 52. Bozkurt et al J Am Coll Cardiol 2003;41:211-4 Criteria Bozkurt et al (n=104) RALES (n=822) NHYA (% patients) I II III IV Undocumented 4.5% 4.5% 15.3% 10.3% 65.4% 0% 0.5% 72% 27% 0% % Patients with LVEF < 35% 54.8% 100% % Patients with renal insufficiency at baseline 30.7% Excluded Use of beta-blockers 34.6% 11%
  • 53.
    • In a highly selected group of patients with severe heart failure who are largely free of other risk factors for hyperkalemia and who can be monitored closely, the addition of spironolactone to standard therapy (as defined in 1994) decreases hospitalization for heart failure and saves lives, with no significant risk of hyperkalemia.
    What RALES said… What we heard…
    • In patients with heart failure, spironolactone saves lives
  • 54. Interpreting Evidence: Key Points
    • Interpretation of evidence depends on:
      • Availability of evidence
      • Ability to understand, assimilate, and interpret evidence and its limitations
  • 55. Putting It All Together
    • What Drives Interpretation?
      • Study design / methodology
        • Appropriate design
        • Inclusiveness / generalizability
      • Statistical analysis and reporting
      • Clinical relevance and ‘sense’
  • 56. Where Does Evidence Fit Into the Decision-Making Process? BELIEFS BEHAVIOUR Evidence Personal Values / Experience Societal Values ‘ Other’ Factors
  • 57. Clinical Research: Writing a Scientific Manuscript
  • 58. What Do Journal Editors Look For? What will maximize the journal’s IMPACT FACTOR? Characteristic Implications Clinical Relevance Research question should be relevant: Ask other clinicians about question’s clinical importance Originality Research question should ideally be unique: Perform a PubMed search Clarity Research question should be simple Study design should be well explained with ONE primary endpoint Before beginning study – know which 4 or 5 figures and/or tables will be produced; make mock figures/tables Know your audience: simple language Good Science Assemble a team: study lead, methodologist, statistician, clinician, at least one senior researcher Must have good design, sufficient sample size, and analysis plan Brevity Keep the manuscript short: 2,000 – 3,000 words Consider different formats: brief reports, research letters
  • 59. Fogg, BJ, et al, What makes Websites Credible? Stanford Persuasive Technology Lab
  • 60. Author Photo Results believable trustworthy competent credible unbiased expert COMPOSITE casual photo no photo formal photo Fogg, BJ, et al, What makes Websites Credible? Stanford Persuasive Technology Lab
  • 61. The Anatomy of a Clinical Research Question
    • Is the question simple?
      • The question must be easy to understand and the results must be easy to convey
      • Start with thinking about the 4 or 5 Tables or Figures that would result from your research
    • Is the question important?
      • It should be important to others besides the researcher
      • Dependent on:
        • Funding source
        • Intended audience
        • Expected impact
  • 62. The Anatomy of a Clinical Research Question
    • Is it the question unique?
      • Must have a solid background / context
      • Extensive literature search to assess novelty of the research
    • Is the question specific?
        • Exposure
        • Outcome: reflects only ONE well-defined primary endpoint
        • Nature of assessment (e.g. relationship vs. difference – need to state intervention)
  • 63. Research Impact
    • High Impact
        • Great questions Great questions
        • Great findings Poor findings
    • Low Impact
    • Poor questions
    • Great findings
    • No Impact
      • Poor questions
      • Poor findings
  • 64. Other Considerations
    • Before beginning your study, know which journal(s) you would like to target
      • Carefully review their format requirements (usually on website)
      • Review sample studies from that journal
    • Focus on a good abstract
      • Many journals screen articles for review based on the abstract
      • Review past article’s abstracts for format and wording style
    • Make cover letter brief but relevant
      • Provide clinical context and rationale for the study
      • Briefly summarize study findings
      • Briefly summarize implications of study findings on clinical practice and/or public health
      • Try to keep it brief – maximum 1 page
  • 65. Research and Communications: An Example
  • 66. Background
    • A particular group of antibiotics called ‘fluoroquinolone’ antibiotics have become the most widely prescribed group of antibiotics
      • About 22 million prescriptions dispensed annually in US for fluoroquinolone antibiotics
      • 16 individual fluoroquinolone antibiotics have been available on the market; In 2005, gatifloxacin (Tequin) was the fluoroquinolone antibiotic of choice for US public health system
    • Some of these drugs have problems: serious adverse events have led to the withdrawal or restriction of several fluoroquinolones in recent years
      • Temafloxacin (blood sugar and kidney problems)
      • Grepafloxacin and sparfloxacin (heart problems)
      • Trovafloxacin (liver problems)
    • In 2005, a student noticed major changes in blood glucose levels in patients receiving Tequin and approached researchers to investigate this observation
  • 67. So We Conducted a Study (Park-Wyllie et al, NEJM, 2006)
    • Population
      • Examined records of over 1.4 million elderly residents of Ontario age 66 years and older between 2002-2004
      • Analysis limited to individuals who were using one of several selected antibiotics
    • Outcomes
      • Hospital admission related to severe changes in blood glucose levels
  • 68. What Were the Results?
    • Hospital admissions related to severe drops in blood glucose levels
      • Gatifloxacin associated with 4x the risk of hospital admission compared to other antibiotics
    • Hospital admissions related to severe increases in blood glucose levels
      • Gatifloxacin associated with almost 17x the risk of hospital admission compared to other antibiotics
  • 69. The Basic Communications Plan
    • Raise the profile of the issue
    • Make the numbers tangible and meaningful
    • Target the right journal
    • Identify the appropriate audiences
  • 70. Making the Numbers Tangible
    • If we consider the 1.4 million elderly residents of Ontario during the study timeframe:
      • There were nearly 17,000 courses of gatifloxacin treatment administered
      • For every 100 courses of gatifloxacin, we may expect 1 hospital admission for dysglycemia
      • On average, it was estimated that at least 1 elderly person in Ontario was hospitalized every week as a result of dysglycemia that was likely associated with gatifloxacin
  • 71. Impact
    • Published early online (March 1 st , 2006) in the New England Journal of Medicine given its clinical relevance
    • Recognized nationally as a significant research contribution:
      • Canadian Society of Clinical Pharmacology
        • Best Publication Award for 2006
    • Gatifloxacin withdrawn from market in May 2006
  • 72.