Pro: The value of randomized controlled studies in dialysis methods

Willis R. Whitney, a chemist, who founded the General Elec-tric Company laboratory once said ‘Necessity is not themother of invention. Knowledge and experiment are itsparents’ [1]. Twardowski and Misra reject the primacy ofexperimentation. They wish us to abandon the necessity todemonstrate by rigorous randomized trials that longer dur-ation and/or more frequent is superior to intermittent dialysis.They argue that randomized controlled studies (RCTs) areinadequate as a research methodology in evaluating a dialysismethod citing limitations with each of the four RCTs—theNational Cooperative Dialysis Study [2], the HEMO study [3],the Frequent Hemodialysis Network (FHN) study [4] and theFHN Nocturnal trial [5]. Twardowski and Misra suggest that‘all progress in dialysis methods was made in research pre-sented in case reports, case control studies and other observa-tional studies. ‘and that’ four RCTs in hemodialysis did notprovide any useful data….observational studies allow clinicalresearch to represent the full breadth of treated patients andoffer tremendous power…’.The approach that Twardowski and Misra is questionable.We can hardly blame rigorous experimentation because ahandful of studies that have been performed so far do notsupport a belief, however strongly held the belief is. After allany given study may have asked the wrong question, or inad-vertently reached the wrong conclusion, or been null becauseof lack of power, or may have problems with study quality(for example, imbalances between the randomized groups oran excessive drop-out rate). Surely, this does not mean thatthe whole methodological approach embodied in RCTs iswrong? RCTs have value in the investigation of interventionslike the choice of dialysis methods, just as they are importantto other forms of human investigation.Twardowski and Misra acknowledge that RCTs representthe most rigorous method to determine whether a cause-and-effect relation exists between any treatment and an outcome.However, they understate the true advantage of RCTs overobservational studies. In the study of interventions, RCTs aresuperior to observational studies because RCTs are based onrandom allocation of subjects to two or more interventiongroups. The possibility of a systematic error is reducedbecause patient-related confounding factors are balancedacross different interventions. While skilled biostatisticiansand epidemiologists might try to attenuate the effect of con-founding, these attempts are generally imperfect and residualconfounding almost always exists. Besides, in RCTs, the inter-vention groups are treated identically, except for the exper-imental treatment. Subjects are analyzed within the group towhich they were allocated, irrespective of whether theyexperience the intended intervention or not (intention totreat analysis) further reducing the chance of bias.Knowledge and experimentation are the founding prin-ciples of evidence-based medicine. The randomized trial rep-resents the most rigorous method to get to the truth ofwhether an intervention causes an outcome. We shouldembrace RCTs not abandon them. RCTs are essential, becauseas Chertow has written [6] ‘Wishing Don’t Make it So’.Evaluating non-randomized studies on dialysismethods: all that glitters is not goldTwardowski and Misra emphasize the importance of non-randomized studies in determining the optimal duration andfrequency of dialysis. These non-randomized studies cover abroad spectrum: case reports, case series, historical controlledtrials and registry data. Collectively, the observational experi-ence has laid the foundation for performing randomizedtrials but is not a substitute.Consider one of the studies cited by Twardowski and Misrain their article to support the superiority of more frequent and/or longer dialysis. This was a study performed by Twardowskibetween 1969 and 1973 and exemplied a ‘before and afterdesign’ [7]. Fourteen dialysis patients were enrolled and after a