Hetastarch Administration in Patients Undergoing Open Heart Surgery in Association with Cardiopulmon

1
Hetastarch Administration in Patients Undergoing
Open Heart Surgery in Association with
Cardiopulmonary Bypass (CPB)

• Blood Products Advisory Committee
• June 14, 2002
• Laurence Landow MD

2
Questions for the Committee

• 1. Is the evidence for excessive bleeding in
  cardiac surgery patients who receive 6
  hetastarch strong enough to warrant a warning
  statement in the hetastarch labeling?
• 2. If there is insufficient evidence for a
  labeling change, should a randomized controlled
  trial(s) be conducted to answer this question?
• If a trial(s) is warranted, please comment on
• Inclusion and exclusion criteria
• What endpoints and differences are clinically
  meaningful
• Major predictors of blood loss

3
(No Transcript)
4
Venous tubing (from the right atrium)
Arterial line microfilter
Arterial tubing (to the aorta)
Roller Pump
5
FDA Background Package

• 5 articles
• 3 retrospective chart reviews
• 1 case-control epidemiology study
• 1 meta-analysis

6
Canver and Nichols

• Chart review
• Use of hetastarch in primary CPB circuitry is
  devoid of any added hemorrhagic risk after
  coronary bypass.

Chest 2000 1181616-1620
7
Knutson et al

• Chart review
• use of hetastarch may increase bleeding and
  transfusion requirements.

Anesth Analg 2000 90801-7
8
Cope et al

• Chart review
• Hetastarch infusion produces a clinically
  important impairment in post-cardiac surgical
  hemostasis.

Ann Thorac Surg 1997 6378-83
9
Herwaldt et al

• Case-control study
• Patient age and hetastarch were risk factors for
  hemorrhage

10
Wilkes et al

• Meta-analysis
• Postoperative blood loss is lower in patients
  exposed to albumin than 6 hetastarch.

Ann Thorac Surg 2001 72527-34
11
Speakers

• Charles C. Canver MD
• Gary R. Haynes MD, PhD
• William Sibbald MD

12
(No Transcript)
13
9 Reasons to Question Conclusions Based on
Non-Randomized Trials
14
Reason 1

• The treatment arms may not be comparable
• Different inclusion and exclusion criteria across
  trials
• Different severity of illness scores (risk
  adjustment)
• Different scoring systems used to assess risk

15
Reason 2

• Even with sophisticated statistical techniques,
  one can never be sure that key outcome predictors
  have been recognized and adjusted for
• Recognized risk factors
• Age, gender, severity of illness
• Unrecognized/unmeasured risk factors
• Genetic predisposition
• Socio-economic status

16
Reason 3

• Standards of medical care change over time
• Canver et al Spanned 8 years

17
Reason 4

• Treatment endpoints vary between protocols
• Knutson et al
• No specific transfusion algorithms used during
  the study period
• No rigorous guidelines for the infusion of
  hetastarch, albumin, or crystalloid

18
Reason 5

• Patient selection and treatment can be biased
• Canver et al perfusionist selected which
  solution to use in the pump prime
• HES might have been avoided in older patients,
  patients with renal failure

19
Reason 6

• Confounding is likely
• Knudson et al HES vs non-HES group
• lower temperatures on bypass
• longer times on bypass
• higher frequency of preoperative anticoagulant
  use

20
Reason 6 (contd)

• Confounding is likely
• Cope et al Volume expansion different across
  groups

21
Reason 6 (contd)

• Confounding is likely
• Pump prime
• Cope et al Albumin crystalloid
• Knutson et al HES not used
• Canver et al HES, crystalloid, albumin, albumin
  HES

22
Reason 7

• Adequate statistical power does not ensure lack
  of bias or confounding
• 200 subjects required to detect an absolute 10
  increase in blood loss
• Knutson et al (N445) confounded with respect to
  CPB time and temperature
• Canver et al (N887) biased with respect to
  patient selection and treatment
• Cope et al (N189) confounded with respect to
  fluid management

23
Reason 8

• The quality of the data is often uneven
• Endpoints
• Different
• Defined differently
• Not pre-specified
• Missing or inaccurate data not easily
  identifiable
• Different variables collected

24
Reason 9

• Reporting bias is possible
• Positive findings are reported whereas negative
  findings are not

25
Conclusion

• Non-randomized clinical trials tend to exaggerate
  effect size

26
Questions for the Committee

• 1. Is the evidence for excessive bleeding in
  cardiac surgery patients who receive 6
  hetastarch strong enough to warrant a warning
  statement in the hetastarch labeling?
• 2. If there is insufficient evidence for a
  labeling change, should a randomized controlled
  trial(s) be conducted to answer this question?
• If a trial(s) is warranted, please comment on
• Inclusion and exclusion criteria
• What endpoints and differences are clinically
  meaningful
• Major predictors of blood loss