Prior studies on reuse-associated mortality have presented conflicting resu
lts and included few adjustments for disease severity or hematocrit levels.
To evaluate the impact of patient and provider characteristics on reuse-as
sociated mortality, we developed a period-prevalent model with a 6-month en
try period. Five cohorts of Medicare hemodialysis patients surviving from J
uly 1 through December 31 of the entry year (1991, 60,985 patients; 1992, 6
3,081 patients; 1993, 76,018 patients; 1994, 82,899 patients; 1995, 91,761
patients) were followed up for the next year. Using a basic Cox regression
survival model (M-1) including age, sex, race, renal diagnosis, prior end-s
tage renal disease time, unit age, unit size, water treatment, dialysate, a
nd germicide, results were compared with those using a more inclusive model
(M-4) adding dialyzer type (conventional or high efficiency/high flux), un
it designation (hospital based or freestanding), unit profit status, comorb
idity, disease severity, and hematocrit. The previous association of for-pr
ofit units with increased mortality was not present after 1994. Whereas the
M-l analysis showed better survival in reuse units after 1991, the more co
mplete M-4 analysis showed no difference in the risk for mortality between
reuse and no-reuse units. We conclude that mortality rates in the United St
ates from 1991 to 1995, when adjusted comprehensively for patient and unit
characteristics, were not different in units that practiced reuse and those
that did not. (C) 2000 by the National Kidney Foundation, Inc.