1. Transfusion Strategies for Patients in Pediatric Intensive Care Units Lacroix J et al. NEJM 2007;356:1609-19 Maggie Constantine, MD, FRCPC Resident, Transfusion Medicine TMR Journal Club May 14, 2007

2. 2 Pediatric ICU RBC Transfusion Objectives  Background  Overview of article  Non-inferiority trial mini-review  Critical appraisal of article

3. 3 Adult ICU RBC Transfusion TRICC Trial  Transfusion Requirements in Critical Care Equivalence trial Liberal vs restrictive transfusion strategies Prospective, randomized Multicenter 1994 to 1997 838 patients  Outcomes Primary  All cause mortality at 30 days Secondary  All cause mortality at 60 days, in-hospital mortality rates NEJM 1999; 340(6)

4. 4 Adult ICU RBC Transfusion TRICC Trial NEJM 1999; 340(6)

5. 5 Adult ICU RBC Transfusion TRICC Trial - Results  Restrictive strategy Average daily Hgb 85 +/- 0.7 g/L 2.6 +/- 4.1 RBC units per day  Liberal strategy Average daily Hgb 107 +/- 0.7 g/L 5.6 +/- 5.3 RBC units per day P=<0.01 NEJM 1999; 340(6)

6. 6 Adult ICU RBC Transfusion TRICC Trial - Results NEJM 1999; 340(6)

7. 7 Adult ICU RBC Transfusion TRICC Trial - Results P=0.10 NEJM 1999; 340(6)

8. 8 Adult ICU RBC Transfusion TRICC Trial - Results NEJM 1999; 340(6)

9. 9 Adult ICU RBC Transfusion TRICC Trial - Conclusions  Results applicable widely  Adhere to transfusion threshold of 70 g/L with a Hgb range of 70 to 90 g/L  Remember those excluded: Active bleeding Chronic anemia Imminent death Pregnancy Admission after a routine cardiac procedure  RBC not pre-storage LRD NEJM 1999; 340(6)

10. 10 RBC transfusions in critically ill patients Background – RCTs Liberal vs Restrictive RBC Strategies StudySetting# patients30-Day Mortality – Liberal [%(n)] 30-Day Mortality – Restrictive [%(n)] Topley et al., 1956Trauma22 Blair et al., 1986GI bleed508.3 (2)0 (0) Fortune et al., 1987Trauma, acute hemorrhage 25 Johnson et al., 1992CVS38 Hebert et al., 1995ICU6925 (9)24 (8) Bush et al., 1997Vascular Surgery998 (4) Carson et al., 1998Ortho (hip #)842.4 (1) Hebert et al., 1999ICU83823.3 (98)18.7 (78) Bracey et al., 1999Cardiac Surgery4282.7 (6)1.4 (3) Lotke et al., 1999Ortho (knee)127 Grover et al., 2006Vascular Surgery260 McIntyre et al., 2006 Trauma (head)671317

11. 11 Adult and Pediatric ICU RBC Transfusion Surveys  Survey of transfusion practices Adult Intensivists Transfusion threshold : 50 to 120 g/L Important transfusion triggers Lactate, low PaO2, shock, age, ER surgery, APACHE II score, chronic anemia, coronary ischemia Pediatric Intensivists Transfusion threshold : 70 to 130 g/L Important transfusion triggers Lactate, low PaO2, active GI bleeding, age, ER surgery, high pediatric mortality score Crit Care Med. 1998 Mar;26(3):482-7. Pediatr Crit Care Med. 2002 Oct;3(4):335-40.

12. 12 The PINT Study - Journal of Pediatrics Sept 2006;149:301-7 Authors’ conclusions  The present findings provide evidence that transfusion thresholds in ELBW infants can be moved downwards by at least 10g/L without incurring a clinically important increase in the risk of death or major neonatal morbidity

13. 13 TRIPICU NEJM 2007; 356(16)

14. 14 TRIPICU Study design  Prospective non-inferiority  19 tertiary-care pediatric ICUs in 4 countries  Treatment arms – pre-storage LRD Restrictive strategy  Transfusion threshold 70 g/L  Target range 85 to 95 g/L Liberal strategy  Transfusion threshold 95 g/L  Target range 110 to 120 g/L

15. 15 TRIPICU Study design  Inclusion Stable, critically ill children  Age 3 days to 14 years  At least one Hgb </= 95 g/L within 7 days after admission to pediatric ICU  Exclusion ICU stay expected to be <24 hours No approval from physician 14 years of age Unstable hemodynamically Acute blood loss Weighed <3 kg Cardiovascular problems Never discharged from NICU Hemolytic anemia Enrolled in another study

16. 16 TRIPICU Study design  Block-randomization  Stratification for center and 3 age groups  Follow-up 28 days Clinical staff and parents were not blinded Statistician and members of the data and safety MC were blinded  Protocol “temporarily suspended” not = to breach of adherence to protocol Acute blood loss Surgical intervention Severe hypoxemia Hemodynamically unstable

17. 17 TRIPICU Study design  Primary outcome Proportion of patients who died during 28 days after randomization, had concurrent MODS or progression of MODS  Secondary outcome Daily PELOD scores, sepsis, transfusion rxns, resp infections, CRI, AE, LOS and death

18. 18 TRIPICU Study design  Statistical design Non-inferiority margin = 10% Need 626 subjects One-sided alpha of 5%; power of 90%  NNT to prevent one red-cell transfusion in RS group  Intention-to-treat and per-protocol analyses

19. 19 TRIPICU Results

20. 20 TRIPICU Results Baseline characteristics similar

21. 21 TRIPICU Results

22. 22 TRIPICU Results

23. 23 TRIPICU Results

24. 24 TRIPICU Results

25. 25 TRIPICU Results – Primary outcomes

26. 26 TRIPICU Results – Secondary outcomes

27. 27 TRIPICU Results – Secondary outcomes

28. 28 TRIPICU Authors’ Conclusions  “…we found that a restrictive transfusion strategy can safely decrease the rate of exposure to red cells as well as the total number of transfusions in critically ill children, even though suspensions of transfusion strategies were permitted under prespecified conditions.”

29. 29 Pediatric ICU RBC Transfusion Non-inferiority trials  Non-inferiority vs. equivalence Impossible to prove two treatments have exact equivalent effects  Inordinately large numbers needed Non-inferiority  Experimental treatment is not worse than an active control by more than the “equivalence margin” Snapinn SM. Curr Control Trials Cardiovasc Med 2000, 1:19-21.

30. 30 Pediatric ICU RBC Transfusion Non-inferiority trials  When might a non-inferiority trial be performed: Applications based upon essential similarity  Modified release products Products with a potential safety benefit over standard When a direct comparison against an active comparator would be acceptable No important loss of efficacy compared to the active comparator would be acceptable Disease areas where use of placebo arm is not possible EMEA/CPMP/EWP/2158/99

31. 31 Pediatric ICU RBC Transfusion Non-inferiority trials  Specifying the non-inferiority margin Specify on the basis of a clinical notion of a minimally important effect  Clearly subjective  Tend to set equivalence margin to be greater than the effect of active control -> harmful treatments fitting within the definition of non-inferiority Specify with reference to the effect of the active control in historical placebo-controlled trials  Historical trials – assumption that effect of active control is similar in trial Snapinn SM. Curr Control Trials Cardiovasc Med 2000, 1:19-21.

32. 32 Pediatric ICU RBC Transfusion Non-inferiority trials  Specifying the non-inferiority margin Generally based not on full effect of active control  Lower bound CI for that effect “fashionable” for non-inferiority margin to be 15%  The smaller the margin the larger the sample size  Per-protocol and intention-to-treat analyses ITT: tends to bias results toward equivalence Per-protocol: can bias results in either direction Both support non-inferiority Snapinn SM. Curr Control Trials Cardiovasc Med 2000, 1:19-21.

33. 33 Pediatric ICU RBC Transfusion Non-inferiority trials  Potential sources of inferiority in non- inferiority trials Selection of patient  Similar population to patient type in whom efficacy of active control has been clearly established Treatment compliance  Also need to document concommitant nonrandomized treatments Outcome measures  Consistent well-defined criteria  Blinding – could give similar scores to both groups Appropriate follow-up Pocock SJ Fundamental & Clin Pharmacol 2003;17:483-490

34. 34 TRIPICU Critical appraisal  Randomization? YES  Were all patients entered into trial properly accounted for? YES  Follow-up complete? NO – protocol violations  Blinding – NOT of patients and clinicians  Were groups similar at start of trial? YES  Concommitant treatments similar? YES

35. 35 TRIPICU Critical appraisal  Treatment effect? Same number of deaths in each arm New or progressive MODS – absolute risk reduction was 0.4% (95% CI, -4.6 to 5.5 with restrictive strategy)  Per-protocol analyses: 0.8% (95% CI, -4.3 to 5.9)  Upper limit of 95% CI did not exceed non-inferiority margin of 10%  ??? 12% in each group… how was “0.4%” calculated  Cannot calculate RR or RRR NNT to prevent one red-cell transfusion was 2 (RS group)  Concerns for this non-inferiority trial  Suspended protocol  Non-blinding of patients and clinicians  Derivation of non-inferiority margin (historical data not referenced)

36. 36 TRIPICU Critical appraisal  Interpretation of conclusion “we found that while a restrictive transfusion strategy decreases the rate of exposure to red cells, it is NO WORSE than a liberal transfusion in terms of MODS in critically ill children” ? Benefit of decreased red cell exposure  Secondary outcomes showed that the restrictive strategy was NO WORSE for  AE  Nosocomial infections  Reactions to RBC  LOS, mechanical ventilation

37. 37 TRIPICU Critical appraisal  Can the results be applied to my patient care? Unclear Unclear how varied a patient population the results can be applied to  4372 excluded for 5399 patients screened  Mindful of exclusion criteria  Suspended protocol: 12% in RS and 6% in LS group Clinically important outcomes considered: YES  Mortality rate exceedingly low  Negative clinical effects resulting from impaired oxygen delivery – MODS  Negative effects of transfusion

38. Transfusion Strategies for Patients in Pediatric Intensive Care Units Lacroix J et al. NEJM 2007;356:1609-19 Maggie Constantine, MD, FRCPC Resident, Transfusion Medicine TMR Journal Club May 14, 2007 Comments? Questions?