Abwägung zwischen Schaden und Nutzen medizinischer Interventionen

Transcription

1 Abwägung zwischen Schaden und Nutzen medizinischer Interventionen Peter Jüni Institut für Sozial and Präventivmedizin, Universität Bern CTU Bern, Inselspital Bern

2 Outline Wall Street, New York, Sept 30, 2004 Rofecoxib: harms benefits Barcelona, Sept 2, 2006 Drug-eluting stents: benefits harms Disclaimer: costs not taken into account

3 Sept 30, 2004

4

5 Merck, Sharp & Dohme: New York Stock Exchange Stock exchange price (US $)

6 Bresalier et al, N Engl J Med 2005

7 Bresalier et al, N Engl J Med 2005

8 FitzGerald, N Engl J Med 2001

9 Arachidonic Acid COX-1 COX 2 Prostaglandins FitzGerald, N Engl J Med 2001

10 Thrombotic risk COX-2 COX-1 Catella-Lawson, Am J Med 2001

11 ( ) trials much larger than those necessary to detect efficacy ( ) in arthritis will be necessary to determine whether cardiovascular consequences ( ) will modulate the anti-inflammatory benefit to be derived from chronic administration of COX-2 inhibitors in humans. McAdam et al, Proc Natl Acad Sci U S A. 1999

12 Bombardier et al, N Engl J Med 2000

13 VIGOR: Ulcer complications 1.0 Cumulative incidence (%) Months Bombardier et al, N Engl J Med 2000

14 2.0 VIGOR: CV events Cumulative incidence (%) Months Mukherjee et al, JAMA 2002 Mukherjee et al, JAMA 2002

15 Relative risk of myocardial infarction: 5.0 (95% CI 1.7 to 14.5) Bombardier et al, N Engl J Med 2000

16 Our results are consistent with the theory that naproxen has a coronary protective effect and Relative risk of myocardial infarction: highlight the fact that rofecoxib does not provide 5.0 (95% CI 1.7 to 14.5) this type of protection owing to its selective inhibition of cyclooxygenase-2 at its therapeutic dose and at higher doses. Bombardier et al, N Engl J Med 2000

17

18

19

20

21 Naproxen s «cardioprotective potential» in VIGOR Rofecoxib s relative risk for MI: 5.0 Naproxen s relative risk for MI: 1/5=0.2 Naproxen s relative risk reduction: %

22 Magnitude of cardio-protective potential -80% -29% -14% Naproxen versus rofecoxib in VIGOR trial Aspirin in RCTs Naproxen in observational studies Relative risk reduction (%)

23 Jüni et al, Lancet 2004

24 Relative risk of myocardial infarction Jüni et al, Lancet 2004

25 RR 2.24 (95% CI 1.24 to 4.02) Relative risk of myocardial infarction Jüni et al, Lancet 2004

26 Bottom line regarding harm Theory: Basic research: RCT vs naproxen: rofecoxib may harm rofecoxib may harm rofecoxib harms RCT vs placebo: rofecoxib harms!

27

28 Observational studies of any help?

29 Papanikolaou et al, CMAJ 2006

30 Relative risk Papanikolaou et al, CMAJ 2006

31 And the benefits?

32

33

34 EDITORIALS Knee Pain Is the Malady Not Osteoarthritis Osteoarthritis is a well-defined pathoanatomic entity readily demonstrable by modern imaging techniques. For a century, the pathology that is this disease has been ingrained in the mind of every medical student April 1992 Annals of Internal Medicine Volume 116 Number 7 Hadler, Ann Intern Med 1992

35 10 cm Visual Analogue Scale Time point: 0 days No Pain Extreme Pain

36 10 cm Visual Analogue Scale Time point: 180 days No Pain Extreme Pain

37 Effect size Difference in pain scores Pooled standard deviation

38 Effect size % 4% Frequency 92% Jüni et al, Best Pract Res Clin Rheumatol 2006

39 Effect size % 7.5% Frequency 85% Jüni et al, Best Pract Res Clin Rheumatol 2006

40 Effect size % 10.5% Frequency 79% Jüni et al, Best Pract Res Clin Rheumatol 2006

41 Effect size % 13.5% Frequency 73% Jüni et al, Best Pract Res Clin Rheumatol 2006

42 Effect size % 16.5% Frequency 67% Jüni et al, Best Pract Res Clin Rheumatol 2006

43 Effect size % 16.5% Frequency 67% Jüni et al, Best Pract Res Clin Rheumatol 2006

44 Effect size % 23.5% Frequency 53% Jüni et al, Best Pract Res Clin Rheumatol 2006

45 Effect size % 31% Frequency 38% Jüni et al, Best Pract Res Clin Rheumatol 2006

46 Effect size % 31% Frequency 38% Jüni et al, Best Pract Res Clin Rheumatol 2006

47 Rofecoxib versus placebo Effect size ~ % 13.5% Frequency 73% Jüni et al, Best Pract Res Clin Rheumatol 2006

48 Rofecoxib versus paracetamol Effect size ~ % 7.5% Frequency 85% Jüni et al, Best Pract Res Clin Rheumatol 2006

49 NNH & NNT Myocardial infarction: NNH ~610 Ulcer complications, versus naproxen: NNT ~130

50 Observational studies of any help?

51 Major cardiovascular disease 42% Peptic ulcer or 3% gastrointestinal bleeding Ray et al, Lancet 2002

52 Baseline Risk and NNT/NNH Number needed to harm Events per 1000 PY

53 Extrapolation to routine settings: rofecoxib & myocardial infarction x 2.24 x 2.24 Control Rofecoxib Events per 1000 PY Trials Routine Jüni et al, Lancet 2005

54 Extrapolation to routine settings: rofecoxib & myocardial infarction x 2.24 x 2.24 Control Rofecoxib Events per 1000 PY NNH 610 Trials NNH 70 Routine Jüni et al, Lancet 2005

55 Risk patterns and serious adverse events overall Rofecoxib versus Naproxen Ulcer complications Thrombotic cardiovascular events Serious adverse events Favours COX-2 Relative risk Favours inhibitors NSAIDs Bombardier et Reichenbach al, N Englet J al, Med ZFA

56 Risk patterns and serious adverse events overall Rofecoxib versus Naproxen Ulcer complications Thrombotic cardiovascular events Serious adverse events Favours COX-2 Relative risk Favours inhibitors NSAIDs Bombardier et Reichenbach al, N Englet J al, Med ZFA

57 Sept 2, 2006

58

59

60 Number of patients Percutaneous coronary interventions in Switzerland Percutaneous coronary interventions Bare-metal stent (BMS) Drug-eluting stent (DES) Year Maeder et al, Kardiovask Med 2007

61 Number of patients Percutaneous coronary interventions in Switzerland Percutaneous coronary interventions Bare-metal stent (BMS) Drug-eluting stent (DES) By end of 2007: ~ patients with drug-eluting stents Year Maeder et al, Kardiovask Med 2007

62 Two FDA approved drug-eluting stents Paclitaxel-eluting stents (PES) Sirolimus-eluting stents (SES)

63 Estimated NNT to avoid one revascularisation over 1 year Bare metal stents versus PTCA NNT 14 Drug-eluting versus bare metal stents NNT 10

64

65

66

67 Lagerqvist et al, N Engl J Med 2007

68 Swedish Registry: overall mortality

69 Swedish Registry: overall mortality RR 1.32 (95%-CI )

70 Swedish Registry: overall mortality RR 1.32 (95%-CI ) NNH 43 (95%-CI )

71 20000 Percutaneous coronary interventions in Switzerland

72 Number of patients Percutaneous coronary interventions in Switzerland Percutaneous coronary interventions Bare-metal stent (BMS) Drug-eluting stent (DES) By end of 2007: ~ patients with drug-eluting stents Year Maeder et al, Kardiovask Med 2007

73 Stettler et al, Lancet 2007

74 38 randomised controlled trials in 18,023 patients Bare Metal Stents 16 comparisons 4,992 patients 7 comparisons 4,312 patients Sirolimus Eluting Stents Paclitaxel Eluting Stents 14 comparisons 7,893 patients

75 Revascularisation

76 Overall mortality

77 Myocardial infarction

78 Bottom line regarding harm Theory: Basic research: RCTs vs BMS: DES may harm DES may harm DES do not harm Observational studies: DES harm Network of RCTs: DES do not harm!

79 Conclusions

80 Magic Bullets

81 No Magic Bullets

82 Balance of benefit and harms We need an understanding of the magnitude of benefits harms in patients who take the drug in clinical routine The balance of benefit and harms will vary according to the spectrum of patients

83 Final common pathways of relevant benefits and harms Overall mortality Serious adverse events Quality-adjusted life-years (?)

84 Extrapolations from Relative risks from RCTs and, for harms, observational studies Event rates observed in routine populations from observational studies

85 To understand the full spectrum of adverse effects those that occur late, that were not known beforehand, and that are rare but nevertheless serious and to be able to investigate the true incidence of known adverse effects in circumstances of actual prescribing, well-designed observational studies will always be necessary. Vandenbroucke & Psaty, JAMA 2008

86 A drug simply cannot be declared safe without measuring the balance of benefits and risks in a randomized controlled trial over an appropriate period of time in a large population representing those who will use the treatment in practice. Califf, JAMA 2005

87 Thank you

88 Vandenbroucke, PLoS Medicine 2008