RCTs, which would then require larger patient populations, would be harder to conduct and more expensive, so that people might be more hesitant about starting them. Also, ethically speaking, more patients would be treated with the placebo when the active product has already been shown to be beneficial. That is why RCTs, such as the PROWESS study, may, and should, be ended before the target number of patients is reached if efficacy is already established at the time of an interim analysis. 2. It is true that RCTs usually include a number of exclusion criteria to increase the chances of finding a beneficial effect. Post-marketing surveillance may be helpful to support the administration of a new intervention in patient populations that were not considered in the initial trial, although the power of such analyses is limited by the lack of a control group [1]. As recently outlined by Califf and DeMets [1], the best thing we can do is to perform additional studies targeting particular patient populations. While the results from any RCT must, of course, be considered in the light of study design, population, and analysis (as must the results from any other study type), the RCT is the best method we currently have to establish the efficacy of therapeutic interventions; denying it would be to deny progress. Placing overly restrictive barriers on RCT development and performance would potentially deprive us of the discov- ery of exciting developments. Indeed, the intensive care community can be proud of its recent achievements in the conduct of good clinical trials [2, 3]. References 1. Califf RM, DeMets DL (2002) Principles from clinical trials relevant to clinical practice. Part I. Circulation 106:1015–1021 2. Van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F, Schetz M, Vlasselaers D, Ferdinande P, Lauwers P, Bouillon R (2001) Intensive insulin therapy in the critically ill patient. N Engl J Med 345:1359–1367 3. Sandham JD, Hull RD, Brant RF, Knox L, Pineo GF, Doig CJ, Laporta DP, Viner S, Passerini L, Devitt H, Kirby A, Jacka M (2003) A randomized, controlled trial of the use of pulmonary-artery catheters in high-risk surgical patients. N Engl J Med 348:5–14 J.-L. Vincent ( ✉ ) Department of Intensive Care Medicine, Erasme University Hospital, Route de Lennik 808, 1070 Brussels, Belgium e-mail: [email protected] Intensive Care Med (2003) 29:859 DOI 10.1007/s00134-003-1717-z CORRESPONDENCE J.-L. Vincent Generalizability of randomized control trials Received: 14 January 2003 Accepted: 20 February 2003 Published online: 8 April 2003 © Springer-Verlag 2003 Sir: I appreciate N. Petrucci’s comments but feel they bring an excessive degree of skepticism about multicenter randomized controlled trials (RCTs). It is indeed true that we treat individual patients rather than patient populations, and it is, therefore, im- portant to define how the results obtained in an RCT apply to any particular patient: 1. I do not agree that we study patients “not to find out anything about them”: The RCT’s primary aim is to serve the population being studied. This is why any patient suffering from a severe disease state should be keen to be en- rolled in an RCT. N. Petrucci suggests that we need to extend our barriers and request a lower statistical p value. I disagree, because such a strategy would only slow down the development of new therapeutic interventions. The

Generalizability of randomized control trials

Download PDF Report

Upload
j-l-vincent
View
215
Download
0

Embed Size (px)

Citation preview

Page 1: Generalizability of randomized control trials

2. It is true that RCTs usually include anumber of exclusion criteria to increasethe chances of finding a beneficial effect.Post-marketing surveillance may behelpful to support the administration of a new intervention in patient popula-tions that were not considered in the initial trial, although the power of suchanalyses is limited by the lack of a con-trol group [1]. As recently outlined byCaliff and DeMets [1], the best thing wecan do is to perform additional studiestargeting particular patient populations.

While the results from any RCT must, ofcourse, be considered in the light of studydesign, population, and analysis (as mustthe results from any other study type), theRCT is the best method we currently haveto establish the efficacy of therapeutic in-terventions; denying it would be to denyprogress. Placing overly restrictive barrierson RCT development and performancewould potentially deprive us of the discov-

ery of exciting developments. Indeed, theintensive care community can be proud ofits recent achievements in the conduct ofgood clinical trials [2, 3].

References

1. Califf RM, DeMets DL (2002) Principles from clinical trials relevant to clinical practice. Part I. Circulation106:1015–1021

2. Van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F,Schetz M, Vlasselaers D, Ferdinande P,Lauwers P, Bouillon R (2001) Intensive insulin therapy in the critically ill patient. N Engl J Med345:1359–1367

3. Sandham JD, Hull RD, Brant RF, Knox L, Pineo GF, Doig CJ, LaportaDP, Viner S, Passerini L, Devitt H, Kirby A, Jacka M (2003) A randomized,controlled trial of the use of pulmonary-artery catheters in high-risksurgical patients. N Engl J Med348:5–14

J.-L. Vincent (✉)Department of Intensive Care Medicine,Erasme University Hospital,Route de Lennik 808, 1070 Brussels, Belgiume-mail: [email protected]

Intensive Care Med (2003) 29:859DOI 10.1007/s00134-003-1717-z C O R R E S P O N D E N C E

J.-L. Vincent

Generalizability of randomized control trials

Received: 14 January 2003Accepted: 20 February 2003Published online: 8 April 2003© Springer-Verlag 2003

Sir: I appreciate N. Petrucci’s commentsbut feel they bring an excessive degree ofskepticism about multicenter randomizedcontrolled trials (RCTs). It is indeed truethat we treat individual patients rather thanpatient populations, and it is, therefore, im-portant to define how the results obtainedin an RCT apply to any particular patient:

1. I do not agree that we study patients“not to find out anything about them”:The RCT’s primary aim is to serve thepopulation being studied. This is whyany patient suffering from a severe disease state should be keen to be en-rolled in an RCT. N. Petrucci suggeststhat we need to extend our barriers and request a lower statistical p value.I disagree, because such a strategywould only slow down the developmentof new therapeutic interventions. The