The HEMO Study: applicability and generalizability

The 2002 Lasker Clinical Medical Research Award honoured Willem Kolff and Belding Scribner, whose seminal work changed kidney failure from a fatal to a treatable disease. What began as exploratory efforts to sustain life and relieve uraemic symptoms in selected patients now provides chronic life-saving replacement therapy to millions of people worldwide. Continued improvements in clinical expertise, accrued scientific information and technical advances have improved dialytic therapy and outcomes. Unfortunately, despite this encouraging trend, there remain significant differences in clinical practice and less than optimal patient outcomes. The annual mortality rate of dialysed patients, albeit variable among different countries, remains unacceptably high. Although medical comorbidities and late referral of patients to nephrologists may contribute to this high mortality rate, the dialytic prescription can also influence mortality. To address this latter concern, national guidelines for dialysis therapy have been developed, but are based, to a great extent, on observational studies and registry data. There have been two randomized clinical trials of the dialysis prescription on outcomes in haemodialysis patients. The first, the National Cooperative Dialysis Study (NCDS), completed >20 years ago, established the importance of dialysis dose in affecting patient outcomes at dose levels substantially below current standards [1,2]. Subsequently, a large number of reports, from observational data sets, suggested that a dose of dialysis substantially higher than that provided in the NCDS trial is associated with a lower mortality rate [3–6]. In response to these observational studies, national standards for dialysis dose were developed in the United States advocating a minimum single pool Kt/V urea (spKt/V) of at least 1.2 [7]. The second randomized clinical trial, the Hemodialysis (HEMO) Study, begun in 1995, was designed to determine whether further increases in dialysis dose above current standards or the use of high-flux membranes would improve patient outcomes. The main results of the HEMO Study, published in December 2002 [8], showed no statistically significant effects by the two study interventions examined, dialysis dose and membrane flux, on mortality rates or on intermediate composite outcomes defined by time to either death or selected first cause-specific hospitalizations. Since the publication of this report, several articles have appeared which discuss the implications of the HEMO Study results for clinical practice, including generalizability, statistical power and design [9,10]. The goal of this report is to address the major questions that have been raised pending the publication of additional analyses of HEMO Study data.