Evidence Based Medicine: Therapy A Critical Look at the PRIMACORP Study

1
Evidence Based Medicine TherapyA Critical Look
at the PRIMACORP Study

• David Goldberg, Eric Silver, Patricia Weng
• Preceptor Walter maestro Eppich

2
Three questions

• Are the results of the study valid?

• What are the results?

• How can I apply the results to my patient?

3
Assessing validity

• Did experimental and control groups begin the
  study with a similar prognosis?
• Were the patients randomized?
• Was randomization concealed (blinded or masked)?
• Were the patients in the treatment and control
  groups similar with respect to known prognostic
  factors?
• Were the patients analyzed in the group to which
  they were randomized?

4
Randomization

• Reduces errors based on treatment bias
• If surgeons only operate on patients who are good
  surgical risks then those who get surgery will
  likely do better than those who do not
• Eliminates bias based on co-morbidities
• If only patients who dont smoke are given
  beta-blockers after an MI but they are compared
  against a pool of untreated patients composed of
  smokers and non-smokers alike, the result of the
  study will not be valid

5
Blinding

• Reduces errors based on patient interpretation
• A patient receiving an experimental therapy may
  feel that they should receive a benefit and thus
  report one when one does not really exist
• Reduces error based on interpretation bias
• If an echocardiographer knows that a patient was
  treated with an afterload reducer, it may affect
  the way he or she estimates the ejection fraction

6
Intention to treat analysis

• Attributing all patients to the group to which
  they were randomized
• Analysis of outcomes based on treatment arm
  rather than treatment actually received
• Preserves value of randomization
• If patients from a group assigned to surgery die
  before they have the surgery, excluding them from
  the group randomized to surgery would skew the
  results by eliminating the patients who are
  potentially the most sick from that treatment arm

7
Assessing the results

• 2 main questions
• How large was the treatment effect?
• How precise was the estimate of the treatment
  effect?

8
Size of treatment effect

• Absolute risk reduction (ARR)
• Difference in the likelihood of a particular
  outcome between the control group (x) and the
  treatment group (y)
• Relative risk (RR)
• Risk of a particular outcome for treatment group
  (y) / risk of the same outcome for the control
  group (x)

9
Size of treatment effect

• Relative Risk Reduction (RRR)
• Equal to (1 Relative Risk) x 100
• A relative risk reduction of 25 means that
  treatment reduces a particular outcome by 25
  relative to the control group
• Number Needed to Treat (NNT)
• Equal to 1 / Absolute Risk Reduction (ARR)
• The number of patients needed to treat in order
  to produce one additional positive outcome or
  prevent one negative outcome

10
Precision of the estimated treatment effect

• 95 confidence interval (CI)
• Related to p lt 0.05
• Range containing true RRR 95 of time
• OR
• If you repeated study 100 times, the result would
  be within the interval 95/100

11
Precision of the estimated treatment effect

• 95 CI
• If N (sample size) is large the 95 CI will be
  narrow and the results more precise
• If the CI includes zero, then the study has not
  demonstrated a significant effect
• This does not mean that it may not show a trend,
  but that the trend has not reached significance
  and the study may need to be repeated with a
  higher power (larger N)

12
Precision of the Estimated Treatment Effect

• With a small N, the RRR may be 25 but the CI
  ranges from -38 to 59 and thus does not
  demonstrate a significant effect
• With a larger N, the CI ranges from 9 to 41 and
  does not include zero leading to the conclusion
  that the therapy does have a significant effect

13
Applying results to patient care

• Were the study patients similar to my own?
• Were all clinically important outcomes
  considered?
• Are the likely treatment benefits worth the
  potential harm and costs?

14
A few caveats
15
When does loss to follow-up threaten validity?
16
A word about endpoints

• Make sure endpoints are relevant and important to
  patients
• Better to look at important outcomes
• Death
• Days in the ICU
• Total number of hospital days
• Days of mechanical ventilation
• Surrogate endpoints
• May not always correlate to clinically important
  outcomes
• May be necessary if other data is unavailable or
  if the study lacks the power to show significance
  for other endpoints

17

• And now to look at the PRIMACORP study in more
  detail, heeeeeeeeerrrrrrrrrrrreeeeees Patty

APPLAUSE
18
Background

• The PRIMACORP study looks at the prophylactic use
  of milrinone in post-operative cardiac patients
  to prevent Low Cardiac Output Syndrome (LCOS) or
  death.

19
Study Organization

• Data Safety Monitoring Board (DSMB) responsible
  for study design and supervision of PRIMACORP
  trial and overseeing safety of the participants
• Clinical Endpoint Committee, a blinded committee,
  mediates and validates the investigators
  evaluation of primary and secondary endpoints

20
Study Design

• Multicenter, randomized, double-blinded,
  placebo-controlled study
• 3 parallel treatment groups in pediatric patients
  undergoing cardiac surgery
• Eligible, stable patients after surgery
  randomized in a 111 ratio within 90 min after
  arrival to ICU
• Stable
• no ongoing cardiopulmonary resuscitation (CPR)
• no initiation of cardiovascular assist device
• no ongoing pulmonary resuscitation that prevents
  placement on a ventilator.

21
(No Transcript)
22
Study Participants

• Inclusion Criteria
• Male and female patient 6 years old who are
  undergoing cardiac surgery involving
  cardiopulmonary bypass (CPB) and who do not have
  preoperative LCOS.
• Parents or legal guardians of all patients must
  give written informed consent before
  randomization.

23
Inclusion criteria eligible surgical procedures

• Ventricular septal defect with
• Interrupted aortic arch
• Arch hypoplasia
• Coarctation of the aorta
• Repair of transposition of the great arteries by
  the arterial switch
• operation
• Complete atrioventricular canal
• Tetralogy of Fallot
• Total anomalous pulmonary venous return
• Truncus arteriosus
• Double outlet right ventricle (biventricular
  repair)
• Anomalous left coronary from the pulmonary artery
• Congenital mitral valve anomaly (may be
  re-operation)
• Aortic valve disease using the Ross operation
  with or without a
• Konno procedure

24
Exclusion Criteria

• Weight 2 kg
• Age 36 weeks postconceptual age
• Creatinine levels gt 1.5 mg/dl within 48 hrs of
  surgery
• LCOS immediately before surgery
• Hypotension immediately prior to study drug
  administration
• Patients currently participating in other
  research studies
• Newborns having surgery for a single ventricle
  causing intercirculatory mixing

25
(No Transcript)
26
(No Transcript)
27
Primary Study Endpoint

• Composite variable consisting of death or
  development of LCOS that required additional
  pharmacological or other support within first 36
  hrs after receiving study drug

28
Secondary Endpoints

• Evaluation of the composite end point of death or
  development of LCOS between 36 hr after receiving
  study drug and the final visit (up to 30 days
  after randomization)
• Duration of mechanical ventilation
• Length of hospital stay
• Total urine output
• Creatinine clearance at the end of study drug
  administration

29
Three questions

• Are the results of the study valid?
• What are the results?
• How can I apply the results to my patient?

Now Eric
30
(No Transcript)
31
Results

• 242 enrolled patients, of which 238 received
  study medication. Due to major protocol
  violations, per-protocol population 227
  patients.
• Age of per-protocol population ranged from 2 days
  to 6.9 years, median 3 months.
• 13 patients received open-label milrinone after
  completion of study drug infusion
• 5 patients lost to follow-up after hospital
  discharge

32

• Using the 227 who received the drug per the study
  protocol, we can calculate the
• - Absolute Risk Reduction (ARR)
• - Relative Risk (RR)
• - Relative Risk Reduction (RRR)
• - Number Needed to Treat (NNT)

33
RRR Relative Risk Reduction
34
Absolute Risk Reduction

• To review
• Absolute risk reduction risk without therapy
  (placebo group) Risk with therapy
• In our study
• - Risk without therapy 26.7
• - Risk with low dose milrinone 17.7
• - Risk with high dose milrinone 9.6
• So, absolute risk reduction
• - Placebo vs low dose 9
• - Placebo vs high dose 17.1

35
Relative Risk Reduction

• To review
• A relative risk reduction of 25 means that
  treatment reduces a particular outcome by 25
  relative to the control group
• Relative risk reduction 1-Relative risk x
  100
• So, first we need to calculate the relative risk,
  which is just
• Risk with therapy
• Baseline risk (placebo)

36
Relative Risk

• Risk with therapy
• Baseline risk (placebo)

Relative risk with low dose (17.7/26.7)
0.66 Relative risk with high dose (9.6/26.7)
0.36
Relative Risk Reduction
Relative risk reduction 1-Relative risk x
100 Relative risk reduction with low dose (1
0.66) X 100 34 Relative risk reduction with
high dose (1 0.36) X 100 64
37
Number Needed to Treat

• The number of patients needed to treat in order
  to produce one additional positive outcome or
  prevent one negative outcome

• 1
• Absolute Risk Reduction

Number Needed to treat with low dose (1/0.09)
11.1 Number Needed to treat with high dose
(1/0.171) 5.8
38
Confidence Interval
RGT x (1 RGT) of control patients
RWT x (1 RWT) of experimental patients

1.96

Risk Without Therapy (RWT), Risk Given Therapy
(RGT)
39
Summary Low dose milrinone vs control
40
Confidence Interval
0
-4.3
22.3
Absolute Risk Reduction () of low dose milrinone
vs control
41
Summary High dose milrinone vs control
42
Confidence Interval
0
5
29.2
Absolute Risk Reduction () of high dose
milrinone vs control
43
Confidence Interval
0
Of note, if youre examining a study and they
dont report the confidence interval for the
Relative Risk Reduction, look at the P value. A P
value of 0.05 means that the lower limit of the
confidence interval is at a Relative Risk
Reduction of 0.
44
Three questions

• Are the results of the study valid?
• What are the results?
• How can I apply the results to my patient?

45
Special thanks to Dr. Walter Eppich for all of
his help
Dr. Walter Eppich