Appendix A Critical Appraisal_Therapy

CAT 2: Therapy
Maribeth Chitkara, MD
Rachel Boykan, MD
Stony Brook Long Island Children’s Hospital
Where Do We Get Evidence?
 Your patients
 Cohort studies
 Colleagues
 Randomized control
 Published anecdotal
trials
 Systematic review
 Meta-analyses
cases
 Case series
Scenario
 4 month old infant w/ RSV bronchiolitis
 Hospitalized for 2.5 days for O2 and IVF
 Family h/o asthma
 Parents wonder if there’s “anything out there” to prevent
the infant from developing asthma
Answerable Clinical Question
PICO
 Patient – Infant w/ RSV bronchiolitis
 Intervention – Given a course of montelukast
 Comparison – Compared with infants without
montelukast
 Outcome – Decrease the subsequent asthma
illnesses?
The Search
 Pubmed search MeSH database
 Terms “Respiratory Syncytial Virus” and
“montelukast”
 You find a double-blind, placebo-controlled study
of infants hospitalized with RSV bronchiolitis
evaluating efficacy of montelukast in reducing
subsequent asthma exacerbations
Bisgaard J for the Study Group on Montelukast and RSV: A Randomized Trial of
Montelukast in RSV Postbronchiolitis. Am J of Respir Crit Care Med 2003; 167:379383.
 Is the study VALID?
 What are the RESULTS?
 Can I APPLY the results to my patient?
Validity

Randomized trial?
 Was Randomization concealed?
 Follow up long enough?
 Patients accounted for at end of trial?
 Were patients, health care workers, data collectors,
and/or data analysts blinded to treatment?
Validity
 Were both groups similar at start of trial?
 Aside from the treatment itself, were both groups
treated equally?
 Did authors utilize an “Intention to Treat”
analysis?
Intention to Treat
 The principle of attributing all patients to the
group to which they were randomized
 Preserves the value of randomization
 Considers the “worst case scenario”
 Scenario 1: Lost experimental subjects fare like
control subjects, lost control subjects fare like
experimental subjects
 Scenario 2: Lost experimental subjects do poorly,
lost control subjects do well
 Scenario 3: Lost subjects in both groups do poorly
Per Protocol Analyses
 Only patients who completed research protocol in
entirety are included in final data analysis
 Excludes all non-compliant patients
 Leaves behind those who may be destined to have a
better outcome
 Has the potential to severely bias the findings in the
study or negate the value of randomization
Therapy – Results
(per protocol)
Outcome
Present
Outcome
Absent
Totals
New
Drug/tx
A
B
A+B
Placebo
C
D
C+D
Totals
A+C
B+D
A+B+
C+D
 Control Event Rate (CER)
 C/C+D
 Experimental Event Rate (EER)
 A/A+ B
 Absolute Risk Reduction (ARR)
 CER - EER
 Relative Risk (RR)
 EER / CER
 Relative Risk Reduction (RRR)
 RRR = [1-RR] x 100
 Number Needed to Treat (NNT)
 1 / ARR
Scenario
RAD No RAD Total Lost to
Follow
UP
 CER = 10/55= 0.18
 EER = 4/61 = 0.07
Montelukast 4
57
61
4
 ARR = 0.18-0.07 = 0.11
 RR = 0.07/0.18 = 0.38
Placebo
10
45
55
Total
14
102
116
10
 RRR = [1-0.38] x 100=62%
 NNT = 1/0.11= 9
Bisgaard J for the Study Group on Montelukast and RSV: A Randomized Trial of Montelukast in RSV
Postbronchiolitis. Am J of Respir Crit Care Med 2003; 167:379-383.
Per Protocol vs ITT
 Per protocol
 ARR = 0.11
 NNT = 9
 ITT (Scenario 1)
 ARR = 0.9
 NNT = 11
 ITT (Scenario 2)
 ARR = 0.3
 NNT = 32
 ITT (Scenario 3)
 ARR = 0.18
 NNT = 5
Number Needed to Treat (NNT)
 Your personal treatment threshold
 I would be willing to treat…
 3 patients to see benefit in one
 5 patients to see benefit in one
 10 patients to see benefit in one
 100 patients to see benefit in one
Results
 How precise was the estimate of treatment effect?
 Calculated results are point estimates
 True result is somewhere within the 95% CONFIDENCE
INTERVAL (CI)
 Tighter the CI, the more likely the calculated numbers are
near the truth
 Larger sample sizelarger the number of outcome
eventsgreater confidence that the true risk is close to
what we have observed
95% Confidence Intervals
for the ARR
 If the value “O” lies in your 95% CI for your ARR, your
result is NOT significant…there may be no difference
between treatment and placebo
 If your 95% CI includes a negative number, your patient
may get worse with treatment…STOP
Applicability
 Are the results applicable to my patient?
 Is our patient so different from those in the study that its
results cannot apply?
 Is the treatment feasible in our setting?
 What are our patient’s potential benefits and harms from
the therapy?
 What are our patient’s values and expectations for both the
outcome we are trying to prevent, and the treatment we are
offering?
Applicability-Scenario
 Do the parents in the scenario have a personal experience
with terrible asthma and would be willing to try anything?
 Is this therapy too expensive for them to consider?
 Does the NNT of 9 influence your or the parents’ decision
about this medication?
 What if the NNT were 2?
Applicability- Bottom Line
 Is this research highly valid, the best there is, and should
influence practice?
 Are there some issues with validity, but the best we have
for now and thus should not dismiss altogether?
 Is it so flawed that we should not consider this new
treatment , or practice without evidence beyond
experience?
Practice Case References
 Case 1: Cohen HA, Varsano I, Kahan E, Sarrell EM, Uziel Y.
Effectiveness of an Herbal Preparation Containing Echinacea,
Propolis, and Vitamin C in Preventing Respiratory Tract Infections
in Children. Arch Pediatr Adolesc Med, 2004; 158:217-221.
 Case 2: Freedman SB, Adler M, Seshadri R, Powell EC. Oral
Ondansetram for Gastroenteritis in a Pediatric Emergency
Department. N Engl J Med, 2006; 354:1698-1705.
 Case 3: Bauchner H, Vinci R, Bak S, Pearson C, Corwin M. Parents
and Procedures: A Randomized Controlled Trial. Pediatrics, 1996;
98:861-867.
EBM References/Resources
 Guyatt, G., Rennie, D., Meade, M.O., & Cook, D.J. (2008)
Users' guides to the medical literature: A manual for
evidence-based clinical practice (2nd ed.). New York:
McGraw-Hill Medical.
 Online EBM Calculator:
http://araw.mede.uic.edu/~alansz/tools.html