RT Journal
A1 Rosansky SJ, Eggers P, Jackson K, Glassock R, Clark WF
T1 SUrvivor bias in early- vs late-start hemodialysis studies—reply
JF Archives of Internal Medicine
JO Archives of Internal Medicine
YR 2011
FD March 14
VO 171
IS 5
SP 477
OP 478
DO 10.1001/archinternmed.2011.66
UL http://dx.doi.org/10.1001/archinternmed.2011.66
AB 
It has been shown in many studies that patients who have more comorbidity and who are older may initiate dialysis earlier than younger patients with less comorbidity.1 Our study used survival modeling to adjust for this confounding and thereby attempted to mathematically make the patients equivalent in early- and late-start dialysis cohorts. This modeling cannot correct for the fact that the patients who remain alive and off dialysis (late starters) may well be a healthier group based on unmeasured factors not associated with the factors already included in the model. Random assignment of patients at some initial time to early and late start is one way of eliminating this issue of survivor bias. Unfortunately, as a recently published randomized controlled trial, The Initiating Dialysis Early and Late (IDEAL) study, demonstrated, even in a research protocol, nephrologists tend to put patients on dialysis at eGFR levels that are commonly accepted by the nephrology community (early-start levels).2 Starting dialysis at lower eGFR levels may be difficult owing to “uremic” symptoms or nephrologists' bias as a result of accepted clinical practice patterns. Thus, we are left with using observational data to decide if early dialysis initiation is beneficial or harmful.