Late Urinary Tract Infection After Renal Transplantation

in the United States

Kevin C. Abbott, MD, S. John Swanson, MD, Erich R. Richter, MD, Erin M. Bohen, MD,
Lawrence Y. Agodoa, MD, Thomas G. Peters, MD, Galen Barbour, MD,
Robert Lipnick, ScD, and David F. Cruess, PhD

● Background: Although urinary tract infection (UTI) occurring late after renal transplantation has been considered
“benign,” this has not been confirmed in a national population of renal transplant recipients. Methods: We
conducted a retrospective cohort study of 28,942 Medicare primary renal transplant recipients in the United States
Renal Data System (USRDS) database from January 1, 1996, through July 31, 2000, assessing Medicare claims for
UTI occurring later than 6 months after transplantation based on International Classification of Diseases, 9th
Revision (ICD-9), codes and using Cox regression to calculate adjusted hazard ratios (AHRs) for time to death and
graft loss (censored for death), respectively. Results: The cumulative incidence of UTI during the first 6 months after
renal transplantation was 17% (equivalent for both men and women), and at 3 years was 60% for women and 47% for
men (P < 0.001 in Cox regression analysis). Late UTI was significantly associated with an increased risk of
subsequent death in Cox regression analysis (P < 0.001; AHR, 2.93; 95% confidence interval [CI], 2.22, 3.85); and
AHR for graft loss was 1.85 (95% CI, 1.29, 2.64). The association of UTI with death persisted after adjusting for
cardiac and other infectious complications, and regardless of whether UTI was assessed as a composite of
outpatient/inpatient claims, primary hospitalized UTI, or solely outpatient UTI. Conclusion: Whether due to a direct
effect or as a marker for serious underlying illness, UTI occurring late after renal transplantation, as coded by
clinicians in the United States, does not portend a benign outcome. Am J Kidney Dis 44:353-362.
This is a US government work. There are no restrictions on its use.

INDEX WORDS: US Renal Data System (USRDS); urinary tract infection; pyelonephritis; death; graft loss; survival;
women; sepsis; creatinine; renal insufficiency; cardiovascular disease; cytomegalovirus.

U RINARY TRACT infection (UTI) can be

associated with significant morbidity after
renal transplantation.1 While the use of postopera-
tive antibiotic prophylaxis has dramatically re-
duced the incidence of UTI after renal transplan-
tation over the past decades,2-4 rates of serious
complications associated with UTI, such as bac-
terial septicemia, remain high for posttransplan-
tation patients even in the modern era.1 However,
late posttransplantation UTI has been widely
considered as “benign,”5,6 based on relatively
small studies.7,8 The majority of transplant cen-
ters administer prophylactic antibiotics posttrans-
plantation,9 but most generally stop antibiotic
prophylaxis 6 months after transplantation.6 How-
ever, emerging concepts from recent studies sug-
gest that UTI, even if late after renal transplanta-
tion, has definite risks,10,11 suggesting that this
clinical entity may not be as “benign” as previ-
ously supposed.
The true impact of posttransplantation UTI
may be further clouded by the circumstances of
death and graft loss occurring among all recipi-
ent groups. Mortality and graft failure are more
common in the first 6 months posttransplanta-
tion, thus potentially skewing the impact of
“early” (ie, within 6 months) versus “late” (ie,
beyond 6 months posttransplantation) UTI. Late
UTI may contribute significantly to subsequent
mortality and graft loss after renal transplanta-
tion. To determine both early and late effects of
posttransplantation UTI on patient survival and
graft loss, we conducted a retrospective cohort
From the Nephrology Service, Walter Reed Army Medical
Center (WRAMC), Washington, DC, and Uniformed Ser-
vices University of the Health Sciences (USUHS), Bethesda,
MD; Organ Transplant Service, WRAMC, Washington, DC,
and National Institutes of Health, Bethesda, MD; NIDDK,
National Institutes of Health, Bethesda, MD; Jacksonville
Transplant Center at Shands, Jacksonville, and the Depart-
ment of Surgery, University of Florida at HSCS, Jackson-
ville, FL; Division of Health Services Administration, Depart-
ment of Preventive Medicine, USUHS, Bethesda, MD;
Division of Epidemiology, Department of Preventive Medi-
cine, USUHS, Bethesda, MD; and Department of Preventive
Medicine and Biometrics, USUHS, Bethesda, MD.
Received March 1, 2004; accepted in revised form April
22, 2004.
The opinions are solely those of the authors and do not
represent an endorsement by the Department of Defense or
the National Institutes of Health.
Address reprint requests to Kevin C. Abbott, LTC, MC,
Nephrology Service, Walter Reed Army Medical Center,
Washington, DC 20307-5001. E-mail: kevin.abbott@
na.amedd.army.mil
This is a US government work. There are no restrictions
on its use.
0272-6386/04/4402-0020$0.00/0
doi:10.1053/j.ajkd.2004.04.040

American Journal of Kidney Diseases, Vol 44, No 2 (August), 2004: pp 353-362 353
354
study of the US renal transplant population from
January 1, 1996, through July 31, 2000. The
objectives were to determine the incidence, risk
factors, and mortality associated with UTI occur-
ring late after renal transplantation (determined
empirically by the association of UTI with sur-
vival after renal transplantation). The null hypoth-
esis was that late UTI is not significantly associ-
ated with subsequent patient or graft survival late
after renal transplantation.
METHODS
Patients
Details of the files used for data abstraction for this study,
as well as limitations of Medicare claim data, have been
described previously and differ by year of selection and
limitations of key variables, notably the use of the Centers
for Medicare/Medicaid Studies Medical Evidence Form
(CMS 2728).12,13 Files used and merged in analysis were
SAF.TXUNOS for the base transplant information, which
was merged with SAF.TXFUUNOS for follow-up informa-
tion, SAF.TXIUNOS for medication information, SAF.PA-
TIENTS for dates and causes of death, SAF.RXHIST60 for
follow-up dates and information on dialysis modality prior
to the date of transplantation, SAF.MEDEVID for informa-
tion on comorbid conditions and laboratory results at the
time of first treatment for end-stage renal disease, and
SAF.INSTITUTIONAL CLAIMS for Medicare claim and
payment information. Physician supplier codes were also
assessed. Consistent with previous reports, only patients
with 2 or more claims for UTI were assessed in analysis.14
Institutional claims were derived from claims for hospital
inpatient stays, hospital outpatient services, most dialysis,
skilled nursing facilities, home health agencies, and hos-
pices. Institutional claims constitute approximately 20% of
total Medicare claims but more than 80% of costs of Medi-
care claims. SAF.PAYHIST was used to verify the dates on
which patients filed Medicare institutional claims with Medi-
care as their primary payer. Files were merged using unique
identifiers. The most recent files released by the United
States Renal Data System (USRDS) include follow-up (in-
cluding dates of death) until October 31, 2001. However, the
most recent dates for Medicare claims available are Decem-
ber 31, 2000. The present study limited analysis to kidney
transplantations occurring in an individual recipient (1 trans-
plant assessed per recipient) with documentation of Medi-
care as primary payer from January 1, 1996, to July 31, 2000
(which could include a repeat transplantation or multiple
organ transplantation).
Outcomes
Our primary dependent (“outcome”) variables were death
from any cause and graft loss (including death with function
as per conventional definitions). Our secondary outcome
was time to the first Medicare claim for UTI.
ABBOTT ET AL
Survival Times
Time to death was defined as time from the date of
transplantation until death, censored for loss to follow-up (as
indicated in USRDS treatment history files) or the end of the
study period, in this case October 31, 2001. Time to graft
loss was defined as the time from the date of transplantation
until the date of graft loss, death, loss to follow-up, or the
end of the study period (October 31, 2001). Time to Medi-
care claims for UTI was primarily calculated as the time
from the date of transplantation until the date of the first
Medicare claim for UTI (whether physician provider or
institutional claim) occurring after renal transplantation dur-
ing the study period, with recipients censored at time of
death, loss to follow-up, 3 years after renal transplantation,
or the end of the study period (December 31, 2000, the most
recent date of Medicare claims available).
Independent Variables
Our primary independent variable was Medicare institu-
tional claims for UTI after renal transplantation. Diagnosis
was based on International Classification of Diseases, 9th
Revision (ICD-9), codes 590.x (“kidney infection,” includ-
ing pyelonephritis both acute and chronic), 595.0x (“acute
cystitis”), or 599.x (“urinary tract infection,” not otherwise
specified). To assess differential effects of subgroups of UTI,
namely pyelonephritis and hospitalization for a primary or
secondary discharge diagnosis of UTI, separate analyses
also assessed whether these outcomes had significant associa-
tions with death and graft loss independent of UTI. “Late”
UTI was defined as UTI occurring after 6 months after
transplantation. UTI was also assessed as a primary hospital-
ization discharge diagnosis and an all-listed diagnosis of the
same ICD-9 codes as above. Patient characteristics and
treatment factors were those at the date of transplantation.
Information on urethral catheterization, postsurgical compli-
cations, antibiotic prophylaxis or treatment, and results of
urinalyses including urine cultures was not available, whereas
cardiovascular risk factors and baseline laboratory values at
the time of presentation to end-stage renal disease (for use in
adjustment of associations with all-cause mortality) were
available for a fraction of patients from CMS Form 2728
(see variables below), obtained at the time of presentation to
dialysis. This information was only available for patients
who presented to dialysis on or after April 1, 1995, or 56% of
the study population. Medicare institutional claims for acute
coronary syndromes (ICD-9 codes 410.x or 411.x), heart
failure (ICD-9 code 428.x), cytomegalovirus disease (ICD-9
code 078.x), and hospitalizations with a primary discharge
diagnosis of sepsis (ICD-9 code 038.x), and all-cause hospi-
talizations of 14 days or more were also used to assess
whether UTI was a marker for sicker patients and thus
confounded by more serious coexisting illness. The duration
of dialysis prior to transplantation was defined as the time
from the first recorded treatment for dialysis therapy until
the date of transplantation. The serum creatinine level was
obtained as the most recent serum creatinine level available
before the end of the first 6 months after renal transplanta-
tion. Other variables assessed included donor and recipient
age, race, sex, weight, body mass index (calculated from
height and weight), induction and maintenance immunosup-
URINARY TRACT INFECTION AFTER RENAL TRANSPLANTATION 355

Table 1. Factors Associated With Medicare Claims for UTI After Renal Transplantation, Renal Transplant
Recipients, January 1, 1996, Through July 31, 2000, With Medicare as Primary Payer

No. of Patients With Factor

Missing (%) Who Had UTI (%)

Total 28,924 12,508 (43.2)

Demographic factors
Male recipient 17,358 (60.0) 0 6,623 (38.2)
Female recipient 11,566 (40.0) 0 5,885 (47.0)*
African-American recipient (v all other races) 7860 (27.2) 0 3,546 (45.1)*
Mean age (y, risk per older year) 45.4 ⫾ 14.6 0 45.9 ⫾ 15.1†
Mean age, men only (y) 46.1 ⫾ 14.4 47.1 ⫾ 14.9†
Mean age, women only (y) 44.7 ⫾ 14.7 44.7 ⫾ 14.5
History of PD (v HD) 9,877 (34.1) 2,850 (28.9)*
Cause of ESRD
Diabetes 7,005 (28.2) 4,220 (14.3) 3,169 (45.2)*
Polycystic kidney disease 1,709 (6.9) 4,220 (14.3) 789 (46.2)*
Chronic pyelonephritis 506 (2.0) 4,220 (14.3) 261 (51.6)*
Chronic obstruction 405 (1.6) 4,220 (14.3) 2,068 (50.9)*
Transplant-related factors
Cadaveric donor (v living donor) 22,445 (77.6) 0 9,953 (44.3)*
Kidney-pancreas 1,009 (3.5) 0 535 (53.0)*
Donor age (y) 36.4 ⫾ 16.1 2,616 (8.8) 36.7 ⫾ 16.6†
Allograft rejection in the first 6 months 3,245 (11.2) Presumed 0 1,353 (41.7)*
posttransplantation
Variables assessed after transplantation
Mean serum creatinine level at 6 months 1.64 ⫾ 1.02 4,702 (16.3) 1.66 ⫾ 1.01†
after transplantation (mg/dL)
Specific cause of hospitalizations at any time
after transplantation
Hospitalizations for heart failure 823 (4.0) Presumed 0 534 (60.1)*
Hospitalizations for ACS 521 (2.5) Presumed 0 315 (47.4)*
Hospitalizations for sepsis 3,915 (13.5) Presumed 0 2,226 (56.9)*
Hospitalizations for CMV 3,071 (10.6) Presumed 0 1,658 (54)*
Comorbidity from CMS form 2728‡
Heart failure 1,872 (11.3) 12,302 (42.5) 849 (45.4)*
Ischemic heart disease 1,262 (7.6) 12,302 (42.5) 536 (42.5)
Serum albumin (g/dL) 3.42 ⫾ 0.68 15,516 (53.6) 3.37 ⫾ 0.68†
Serum creatinine (mg/dL) 9.76 ⫾ 4.42 12,536 (43.3) 9.26 ⫾ 4.21†
Hematocrit (%) 28.13 ⫾ 5.7 13,416 (46.4) 28.21 ⫾ 5.7

NOTE. Data are given as the number (% of total) or mean ⫾ 1 SD. Dates for renal transplant recipients were January 1,
1996, through July 31, 2000, censored at 3 years of follow-up. To convert creatinine levels in mg/dL to ␮mol/L, multiply by
88.4; serum albumin levels in g/dL to g/L, multiply by 10.
Abbreviations: PD, history of peritoneal dialysis for any 60-day period before transplant (no history of peritoneal dialysis);
HD, heart disease; ACS, acute coronary syndrome; CMV, cytomegalovirus.
*P ⬍ 0.05 versus patients without UTI by chi-square test.
†P ⬍ 0.05 versus patients without UTI by Student’s t-test.
‡Obtained only for patients who presented to dialysis on or after April 1, 1995, so limited to 16,662 (57.6%) patients.

pressive medications (specifically use of mycophenolate
mofetil), graft loss, previous transplantation, delayed graft
function, results of serologic testing for cytomegalovirus
(both donor and recipient), donor kidney positive for hepati-
tis C virus, end transplant center, and allograft rejection in
accordance with previous studies. Specific causes of end-
stage renal disease assessed included diabetes mellitus,
chronic pyelonephritis, chronic obstruction, polycystic kid-
ney disease, and systemic lupus erythematosus (other causes
were not excluded but were not specifically identified).
History of donor UTI was available only for recipients of
cadaveric donors, and therefore analysis using this variable
was restricted to cadaver donors. Treatment with peritoneal
dialysis for any 60-day period prior to transplantation was
obtained from patient treatment files. Data from CMS Form
2728 was available for over half of the cohort (Table 1)
whose first date of end-stage renal disease was on or after
April 1, 1995, due to time elapsed from presentation to
end-stage renal disease until renal transplantation, therefore
disproportionately including recipients of living donor kid-
356
neys. This form included information on pertinent labora-
tory values and baseline comorbidities, predominantly car-
diovascular disease.
Statistical Analysis
All analyses were performed using SPSS 12.0 (SPSS,
Inc., Chicago, IL). Files were merged and converted to SPSS
files using DBMS/Copy (Conceptual Software, Houston,
TX). Additional analyses were validated in Stata 8.0, Inter-
cooled (College Station, TX). Analysis was performed with
chi-square testing for categorical variables (Fisher’s exact
test was used for violations of Cochran’s assumptions) and
Student’s t-test for continuous variables (Mann-Whitney test
was used for non-normally distributed variables), respec-
tively. Statistical significance for univariate comparisons
was defined as P ⬍ 0.05. Variables with P ⬍ 0.10 in
univariate analysis for a relationship with development of a
first Medicare claim for UTI were entered into multivariate
analysis as covariates, because of the possibility of negative
confounding. Variables thought to have a known clinical
reason to be associated with UTI were introduced into
multivariate models even if univariate P values were greater
than 0.10, in accordance with established epidemiologic
principles.15 Continuous variables were explored and values
thought to be inconsistent with clinical experience were
excluded.
The independent association between patient factors and
time to the first Medicare claim for UTI during the study
period was examined using multivariate analysis with Cox
regression, (likelihood ratio method16), controlling for vari-
ables entered into the model as described above. Both formal
and graphic assessments for all covariates in the final models
were performed to verify the existence of proportional
hazards. Multivariate analysis will exclude all patients with
missing values, likely resulting in substantially smaller mod-
els than the entire study population. Continuous variables
that were non-normally distributed were also assessed by
quartiles. Models also substituted values for missing values
of variables (“interpolation”) to assess for potential bias
resulting from exclusion of patients with missing values.
Hierarchically well-formed models were used for the assess-
ment of interaction terms. Interaction terms that were found
to be statistically significant in Cox regression analysis were
presented stratified for each level of the covariate, as appro-
priate.17 The independent association of Medicare claims for
UTI with patient mortality and graft loss was determined
using Cox nonproportional hazards regression assessing the
time of the first Medicare claim for UTI as a time-dependent
variable, with times after the first Medicare claim for UTI
coded as 1, and all other times coded as 0, as in previous
reports.18,19 Other hospitalizations (acute coronary syn-
drome, heart failure, cytomegalovirus, and sepsis as men-
tioned above) were also assessed as time-dependent vari-
ables in analysis. To assess for possible bias of repeated UTI
affecting survival associated with late UTI, analysis of late
UTI also excluded patients who were coded for UTI during
the earlier period of the study. In addition, a separate
variable coding for multiple (2 or more) institutional claims
for UTI were assessed in analysis. Analysis of the associa-
tion between UTI and outcomes (death and graft loss) is
further complicated by the fact that rates of UTI, death, and
ABBOTT ET AL
graft loss are not constant over time; in fact, rates of all 3
outcomes are more common early after transplantation than
late. Therefore, late UTI was defined as UTI occurring 6
months or more after renal transplantation, and outcomes
assessed were death or graft loss occurring after the diagno-
sis of UTI, compared with recipients who did not develop
UTI. Because pyelonephritis might have different implica-
tions than other classifications of UTI, analysis was also
performed separately using late pyelonephritis (ICD-9 code
590.x) as an outcome. Since approximately half of Medicare
claims for UTI were institutional (hospitalization claims),
specific-cause hospitalizations were assessed as time-
dependent variables in the final model. Also, analysis was
performed excluding patients who had been hospitalized for
heart failure, acute coronary syndrome, sepsis, or cytomega-
lovirus. Analysis also was stratified by sex and by receipt of
kidney-pancreas transplant, given previous reports that blad-
der drainage of pancreas allografts may be associated with a
higher risk of UTI.20,21
RESULTS
Of 59,077 recipients of renal transplants from
January 1, 1996, to July 31, 2000, 29,597 had
valid follow-up times and evidence of Medicare
as primary payer at the time of transplantation.
Of these, 28,942 (97.7%) had Medicare payment
dates with valid Medicare as primary payer sta-
tus (by SAF.PAYHIST) within 14 days of trans-
plantation. During the study period, 12,508
(43.2%) had either an inpatient or outpatient
claim for UTI (composite UTI); 12,803 recipi-
ents (44.3%) had physician supplier (outpatient)
claims for UTI, 8,812 (28.0%) had institutional
(inpatient) claims for UTI, and 6,124 (21.2%)
had both inpatient and outpatient claims for UTI.
All subsequent results refer either to composite
UTI (as defined), UTI as an inpatient all-listed
diagnosis (institutional claims), or UTI as a pri-
mary hospital discharge diagnosis. As shown in
Fig 1, the incidence of composite UTI during the
first 6 months after renal transplantation was
17% (equivalent for both men and women), and
at 3 years was 60% for women and 47% for men
(P ⬍ 0.001 by log rank test). Almost all patients
with outpatient claims for UTI had multiple
claims. Among patients with institutional claims
for UTI (all-listed diagnoses), 62.9% had a single
claim for UTI, 18.8% had 2 institutional claims,
7.5% had 3 institutional claims, with the remain-
der having 4 or more institutional claims for
UTI. Of recipients with Medicare institutional
claims for later UTI, 85% were coded as UTI
(not otherwise specified, 599.0x), 13.0% as pyelo-
nephritis (590.1x), and 1.3% as cystitis. There
URINARY TRACT INFECTION AFTER RENAL TRANSPLANTATION 357

Fig 1. Time to first Medi-

care claim for UTI (all-listed
diagnoses, as defined in the
Methods section) after renal
transplantation, by sex, US
renal transplant recipients,
1996-2000. The peak risk of
UTI was in the first 6 months
after transplantation, with no
difference by sex until after 6
months when female recipi-
ents were at significantly
greater risk of UTI (P < 0.01
by log rank test); also signifi-
cant in multivariate analysis
(see Table 2).

was no difference in site of UTI by sex. Among
patients with UTI coded as a primary discharge
diagnosis (n ⫽ 2,225, 7.7%), the leading diagno-
sis was UTI unspecified (599.0x, 69%) and acute
pyelonephritis (23%). Among women, 24% of
primary discharge diagnoses were for acute py-
elonephritis versus 19% in men (P ⬍ 0.01 by
chi-square test). Very few (0.8%) were coded as
cystitis.
Of the 8,812 recipients with Medicare institu-
tional claims for UTI, 4,863 (59.9%) occurred as
a primary or secondary hospitalization discharge
diagnosis; of these, 2,225 (25.6%) were for a
primary discharge diagnosis. A higher proportion
of patients with UTI occurring within the first 6
months had UTI coded as a primary discharge
diagnosis than UTI occurring more than 6 months
after transplantation (46.7% versus 19.6%, P ⬍
0.001 by chi-square test). Among patients with a
primary or secondary discharge diagnosis of UTI,
the most common primary discharge diagnosis
was code 996.81, complication of kidney trans-
plant (21.7%), followed by code 599.x, UTI
unspecified (15.9%).
The time to UTI, stratified by recipient sex, is
shown as a Kaplan-Meier plot in Fig 1. The peak
risk of UTI occurred in the first 6 months after
renal transplantation, and the risk of UTI was the
same for men and women. After this, the risk of
UTI was significantly higher for female recipi-
ents than for male recipients.
Patient characteristics associated with compos-
ite UTI are shown in Table 1. Female sex, African-
American race, older age in men only, history of
peritoneal dialysis (versus hemodialysis), and
cause of end-stage renal disease (diabetes, poly-
cystic kidney disease, chronic pyelonephritis,
and chronic obstruction), cadaver kidney, kidney-
pancreas transplant, older donor age, rejection in
the first 6 months, and elevated serum creatinine
within 6 months after transplantation were signifi-
cantly associated with UTI. Specific hospitaliza-
tions as shown were associated with UTI. In
information from CMS Form 2728, heart failure,
lower serum albumin, and lower serum creati-
nine were associated with increased risk of UTI.
Table 2 shows results of Cox regression analy-
sis of factors independently associated with a
shorter time to UTI. A high serum creatinine
level, as well as recipient female sex, longer
duration of pretransplantation diabetes, older re-
cipient age for men, renal failure due to diabetes,
chronic pyelonephritis, chronic obstruction, poly-
cystic kidney disease, and kidney-pancreas trans-
plantation were independently associated with
an increased risk of later UTI. Among data from
CMS Form 2728, prevalent heart failure (but not
ischemic heart disease), lower serum albumin,
and lower serum creatinine at the time of dialysis
were significantly associated with a higher risk
of later UTI. Risk factors for multiple episodes
358 ABBOTT ET AL

Table 2. Cox Regression Analysis of Factors Independently Associated With New Medicare Claims for
Composite UTI Occurring 6 Months After Renal Transplantation

Adjusted Rate Ratio for UTI

P Value in Cox Regression*

Serum creatinine level 6 months after transplantation (mg/dL) ⬍0.001 1.09 (1.06,1.13)
Female recipient (v male) ⬍0.001 1.86 (1.76,1.97)
Age (in years, limited to men only), ⬎55.4 v lower ⬍0.001 1.36 (1.21,1.54)
Quartiles of years of dialysis before transplantation (v ⬍ 1.00) 0.012 1.25 (1.13,1.37)
⬎2.98 v less
Cause of ESRD (v all others)
Diabetes ⬍0.001 1.19 (1.11,1.33)
Chronic pyelonephritis 0.002 1.46 (1.24,1.72)
Chronic obstruction ⬍0.001 1.54 (1.26,1.89)
PKD 0.007 1.23 (1.04,1.29)
Kidney-pancreas transplant 0.018 1.18 (1.03,1.35)
N in final model 17,395
Model also including data from CMS Form 2728*
Prevalent congestive heart failure ⬍0.001 1.26 (1.11,1.43)
Serum albumin ⬍3.0 mg/dL v higher 0.001 1.26 (1.10,1.43)
N in final model 6,730

NOTE. Data are given as the number (% of total) or mean ⫾ 1 SD. Dates for renal transplant recipients were January 1,
1996, through July 31, 2000, censored at 3 years of follow-up. Patients who died, were lost to follow-up, or reached the end of
the study period within 6 months after transplantation were not included in analysis. To convert creatinine in mg/dL to ␮mol/L,
multiply by 88.4; albumin in g/dL to g/L, multiply by 10.
Abbreviations: ESRD, end-stage renal disease; PKD, polycystic kidney disease.
*Obtained only for patients who presented to dialysis on or after April 1, 1995, as per Table 1 (57.6%) patients. In this
model, all factors presented above in the larger model were still significant with similar point estimates except for diabetes,
which was no longer significant, AHR for UTI, 1.01; 95% CI, 0.91, 1.13; P ⫽ 0.86.

of UTI (2 or more Medicare institutional claims
for UTI) were similar; that is, risk factors for
multiple episodes of UTI were similar to risk
factors for single UTI episodes. Results were
similar using interpolated values for missing
values of variables.
Cumulative patient survival after UTI as a
composite diagnosis was 87% at 1 year, 81% at 2
years, and 76% at 3 years. Cumulative graft
survival after UTI was 84% at 1 year, 78% at 2
years, and 73% at 3 years. Figure 2 shows
adjusted hazard ratios for the association be-
tween late UTI, death, and graft loss, respec-
tively. UTI was assessed both as composite UTI
(inpatient and outpatient claims, all-listed diag-
noses) and hospitalizations with primary dis-
charge diagnoses of UTI (primary hospitalized
UTI). Variables fitted in the final Cox regression
model are specified in the legends to the figures.
As shown, composite late UTI was indepen-
dently associated with an increased risk of subse-
quent death, adjusted for other factors, with no
interaction between UTI and other factors for the
association with either death or graft loss. UTI as
a primary discharge diagnosis had a statistically
significant interaction with sex in the association
with death; therefore, adjusted hazard ratios are
shown separately for men (in whom UTI as a
primary discharge diagnosis was independently
associated with death) and for women (in whom
this association was not significant). A variable
coding for the occurrence of multiple (more than
1 Medicare institutional claim) for UTI was not
independently significant in analysis, either over-
all or separately for men or women; that is,
single episodes of UTI were associated with
mortality to a similar degree as multiple episodes
of UTI.
Additional analysis was performed with UTI
defined as an outpatient diagnosis only occurring
more than 6 months after transplantation (from
physician supplier codes, N with outpatient UTI
only ⫽ 4,302, 19.1%), excluding patients who
had inpatients claims for UTI (total N in model ⫽
13,041). In this analysis, even outpatient UTI
was significantly associated with subsequent risk
of death using the same covariates as above
(adjusted hazard ratio [AHR], 1.33; 95% confi-
URINARY TRACT INFECTION AFTER RENAL TRANSPLANTATION 359

Fig 2. Graphic representa-

tion of the AHR of mortality for
Medicare claims for UTI (all-
listed diagnoses) assessed as
a time-dependent variable with
subsequent death (A) and graft
loss (B), respectively. UTI was
assessed as late (occurring
more than 6 months after trans-
plantation) both as a compos-
ite of inpatient and outpatient
claims and as a primary hos-
pital discharge diagnosis.
Whether as a composite diag-
nosis or as a primary dis-
charge diagnosis, late UTI was
independently associated with
subsequent risk of death. Co-
variates included donor and re-
cipient age, race, sex, donor
type, history of rejection in the
first 6 months, most recent
serum creatinine level at 6
months, history of heart fail-
ure, ischemic heart disease,
smoking, body mass index,
serum albumin, diabetes, my-
cophenolate mofetil use at
discharge, duration of dialy-
sis prior to transplantation,
and hospitalization for heart
failure, cytomegalovirus, or
sepsis as time-dependent
variables. Additionally, mod-
els for graft loss adjusted for
human leukocyte antigen mis-
match, patient sensitization,
donor race, and donor type.
*P < 0.05 by Cox regression.

dence interval [CI], 1.15, 1.12; P ⬍ 0.001) with
no interaction between UTI and sex. Outpatient
UTI was also significantly associated with graft
loss (AHR, 2.35; 95% CI, 1.61, 3.43; P ⬍ 0.001)
excluding patients with inpatient graft loss.
Separate analysis of kidney-pancreas recipi-
ents revealed that late UTI was also indepen-
dently associated with death (AHR, 4.06; 95%
CI, 1.55, 10.63); the association of UTI with
graft loss was not significant (AHR, 1.21; 95%
CI, 0.19, 7.81)
There were no substantial differences in speci-
fied causes of graft loss between patients with
UTI and those without UTI. Specifically, among
patients with specified causes of graft loss (ie,
excluding patients with missing or unknown
causes or patients who died with functioning
grafts), infection was listed as a cause of graft
loss in 6.5% of patients with UTI versus 5.0% of
those without UTI (not significant), and chronic
rejection accounted for over 50% of specified
causes of graft loss for both groups, followed by
acute rejection, accounting for 16% and 19%,
respectively. As shown in Figure 2, both late
composite UTI and primary hospitalized UTI
after renal transplant were significantly associ-
ated with an increased risk of graft loss, P ⬍
0.001. There was no significant interaction be-
tween UTI and sex for the risk of graft loss,
either early or late. Among patients with UTI
who died, primary specific causes of death were
missing or unknown for 61%. Cardiovascular
death was coded less commonly as a primary
cause of death among patients with total UTI
(20.7% versus 23.9% for those without UTI)
while coding for infectious death was no differ-
ent (12.6% versus 12.5%). This association did
not differ by sex.
360
DISCUSSION
In the present study of the USRDS renal
transplant population, the novel finding was that
UTI occurring late after renal transplantation,
regardless of how defined, was independently
associated with an increased risk of subsequent
recipient death and graft loss. The idea that late
UTI after renal transplantation is “benign” de-
rives mainly from 2 relatively small studies.7,8
Further, recent reports suggest that many patients
with late UTI present with advanced infec-
tions.10,11 Because causes of death were not sig-
nificantly different by sex among patients with
UTI (either as all-cause or primary), the finding
of an interaction between primary hospitalized
UTI and sex with subsequent risk of death should
be interpreted with caution. One possible expla-
nation for this finding is pseudointeraction result-
ing from misclassification of UTI,22 because many
patients who actually had UTI would be coded as
not having UTI according to this more restrictive
definition. In particular, UTI as a cause of sepsis,
certainly a serious clinical circumstance, is often
listed in the second discharge diagnosis position,
after sepsis.1 However, we previously reported
that female renal transplant recipients were more
likely to have sepsis due to enteric gram-
negative organisms and had a lower risk of death
associated with sepsis than men; men had a
higher risk of sepsis due to Pseudomonas sp and
staphylococcal infections.1 Thus, this interaction
may also reflect differences in the spectrum of
UTI between sexes. We were unable to deter-
mine the causative agent of UTI in this analysis.
In any case, the association of UTI with death
persisted after adjusting for cardiac and other
infectious complications, and regardless of
whether UTI was assessed as a composite of
outpatient/inpatient claims, primary hospitalized
UTI, or solely outpatient UTI.
The leading cause of death after UTI in the
present study was cardiovascular disease, the
most single specific cause cited of which was
“cardiac arrest, cause unknown,” and autopsy
verification could not be determined. Causes of
death were not reliable enough in the database
for statistical analysis, and infectious causes of
graft loss were not substantially higher in pa-
tients with UTI. However, the leading cause of
death in renal transplant recipients after hospital-
ABBOTT ET AL
ized septicemia was also cardiovascular disease,
and the majority of these deaths occurred after
hospital discharge,1 casting some doubt as to
whether all these deaths are truly cardiovascular
related and also whether persistent inflammation
contributes to cardiovascular as well as infec-
tious death. A similar relationship between septi-
cemia and causes of death was noted among
long-term dialysis patients,23 reinforcing the as-
sociation between persistent inflammation and
cardiovascular disease in this population.24 The
association of posttransplantation renal insuffi-
ciency, which was commonly associated with
UTI in this analysis, with cardiovascular compli-
cations after renal transplantation is well estab-
lished.25,26
We found that both all-cause UTI and primary
UTI were significantly associated with graft loss.
Two recent cohort studies have yielded conflict-
ing results on the association of the timing of
UTI and graft loss. Muller et al found that
patients who experienced chronic allograft ne-
phropathy had a significantly higher rate of late
UTI (more than 2 years after transplantation)
than patients who did not experience chronic
allograft nephropathy.27 In contrast, Giral et al
found that early, but not late, graft pyelonephritis
was significantly associated with increased rates
of graft loss.28 However, graft survival in general
was higher in that cohort than reported in the
USRDS population, and thus results may not be
generalizable to US recipients. Further, the
present study assessed the association of late
UTI with subsequent graft loss compared with
comparable recipients whose time at risk started
at more than 6 months after transplantation, with
substantially larger sample size than previous
studies. With respect to specific causes of graft
loss, Matas et al reported that among cadaveric
kidney recipients, the composite of “acute rejec-
tion/infection” accounted for 3.8% of causes of
graft loss including death with function.29 When
death with function was excluded (as we did in
the present analysis), this percentage changed to
5.6% (acute rejection was also listed separately
as a cause of graft loss in that study, introducing
the possibility of redundant or ambiguous cod-
ing), not inconsistent with results of the present
study.
While a history of diabetes, chronic infection
or obstruction, or polycystic kidney disease (as
URINARY TRACT INFECTION AFTER RENAL TRANSPLANTATION
causes of renal failure) certainly predisposed to
later UTI, as indicated in Table 2, both our data
and the cited studies indicate that most late
recipient UTIs occur independent of these risk
factors. It was remarkable that the risk of UTI
was the same for either sex during the first 6
months after transplantation. In fact, no specific
risk factors were significantly associated with
early UTI, a time when almost all patients have
urethral catheters in a setting of their transplant
surgeries. Although the present study could not
determine use of prophylactic antibiotics, most
transplant centers prescribe prophylactic antibiot-
ics to their recipients for the first 3 to 6 months
after transplantation.9
The key potential limitation of this study is
that UTI, as we have assessed it, may merely be a
marker for sicker patients and not itself cause
either a higher risk of death or graft loss. This
limitation may apply to either UTI assessed as an
all-listed (total) diagnosis or as a primary dis-
charge diagnosis. UTI as an all-listed diagnosis,
particularly in hospitalized patients, could reflect
patients admitted for other conditions who devel-
oped nosocomial UTI, and the subsequent risk of
adverse outcomes could be related to the admit-
ting diagnosis rather than UTI. Whether this
limitation applied to outpatient cases of UTI
could not be determined. Further, the prioritiza-
tion of the diagnosis of hospitalized UTI may not
necessarily reflect the severity of the diagnosis.
For example, in our previous analysis of the
USRDS renal transplant population, we found
that UTI was the leading secondary diagnosis for
patients with a primary discharge diagnosis of
sepsis1; ie, sepsis attributed to UTI. Therefore,
coding of UTI as a secondary discharge diagno-
sis does not necessarily mean that these UTIs
were less serious than “primary” UTIs, given
limitations of coding data. We adjusted for other
serious conditions to the extent possible from
registry data, and in addition assessed UTI as a
composite outpatient or inpatient diagnosis, re-
flecting the full spectrum of exposure as in previ-
ous analysis of the USRDS.14 Other potential
confounders were prolonged hospitalization, post-
transplantation renal insufficiency/graft dysfunc-
tion, and cardiovascular disease. However,
whether as all-cause or as a primary diagnosis,
UTI was independently associated with adverse
outcomes adjusted for these factors. Other risk
361
factors associated with UTI in the present study
were similar to risk factors for UTI in the general
population and did not reflect risk factors for
cardiovascular disease. Medicare claims do not
reliably capture medication use, and therefore
antibiotic use could not be determined. We were
unable to determine the role of postoperative
complications, such as urine leak or lymphocele
formation, or cumulative immunosuppression
dose on the future risk of UTI. Presumably all
renal transplant recipients had urethral catheter-
ization at their initial surgery, but we were unable
to determine the duration of urethral catheteriza-
tion. We were also unable to determine the pres-
ence of persistent anatomic abnormalities (other
than as specified causes of renal failure), which
could have played a role in late UTI risk.
In conclusion, the findings of the present study
of the US renal transplant population contrast
with previous studies in that late UTI was inde-
pendently associated with both an increased risk
of subsequent death and graft loss. Whether this
is because late posttransplantation UTI truly con-
tributes to an increased risk of death and graft
loss or is a marker for more serious conditions,
such as posttransplantation renal insufficiency or
cardiovascular disease, could not be conclu-
sively determined due to the observational nature
of the database and our inability to confirm
diagnoses. However, in either case, it is clear that
hospital-associated UTI occurring late after renal
transplantation, as coded by clinicians in the
United States, does not portend a benign out-
come. Follow-up studies to assess the causes of
the increased risk of death and graft loss associ-
ated with UTI in this population could shed light
on issues not fully discernable from the USRDS.
REFERENCES
1. Abbott KC, Oliver JD 3rd, Hypolite I, et al: Hospital-
izations for bacterial septicemia after renal transplantation in
the United States. Am J Nephrol 21:120-127, 2001
2. Cohen J, Rees AJ, Williams G: A prospective random-
ized controlled trial of perioperative antibiotic prophylaxis
in renal transplantation. J Hosp Infect 11:357-363, 1988
3. Fox BC, Sollinger HW, Belzer FO, et al: A prospec-
tive, randomized, double-blind study of trimethoprim-
sulfamethoxazole for prophylaxis of infection in renal trans-
plantation: Clinical efficacy, absorption of trimethoprim-
sulfamethoxazole, effects on the microflora, and the cost-
benefit of prophylaxis. Am J Med 89:255-274, 1990
4. Hibberd PL, Tolkoff-Rubin NE, Doran M, et al: Tri-
methoprim-sulfamethoxazole compared with ciprofloxacin
362
for the prevention of urinary tract infection in renal trans-
plant recipients. A double-blind, randomized controlled trial.
Online J Curr Clin Trials Doc No 15, 1992
5. Schmaldienst S, Dittrich E, Horl WH: Urinary tract
infections after renal transplantation. Curr Opin Urol 12:125-
130, 2002
6. Brown PD: Urinary tract infections in renal transplant
recipients. Curr Infect Dis Rep 4:525-528, 2002
7. Rao KV, Andersen RC: Long-term results and compli-
cations in renal transplant recipients. Observations in the
second decade. Transplantation 45:45-52, 1988
8. Grekas D, Thanos V, Dioudis C, et al: Treatment of
urinary tract infections with ciprofloxacin after renal trans-
plantation. Int J Clin Pharmacol Ther Toxicol 31:309-311,
1993
9. Midtvedt K, Hartmann A, Midtvedt T, et al: Routine
perioperative antibiotic prophylaxis in renal transplantation.
Nephrol Dial Transplant 13:1637-1641, 1998
10. Oguz Y, Bulucu F, Oktenli C, et al: Infectious compli-
cations in 135 Turkish renal transplant patients. Cent Eur J
Public Health 10:153-156, 2002
11. Ghasemian SM, Guleria AS, Khawand NY, et al:
Diagnosis and management of the urologic complications of
renal transplantation. Clin Transplant 10:218-223, 1996
12. Longenecker JC, Coresh J, Klag MJ, et al: Validation
of comorbid conditions on the end-stage kidney disease
medical evidence report: The CHOICE study. Choices for
Healthy Outcomes in Caring for ESRD. J Am Soc Nephrol
11:520-529, 2000
13. Abbott KC, Bucci JR, Cruess D, et al: Graft loss and
acute coronary syndromes after renal transplantation in the
United States. J Am Soc Nephrol 13:2560-2569, 2002
14. Kasiske BL, Snyder JJ, Gilbertson D, et al: Diabetes
mellitus after kidney transplantation in the United States.
Am J Transplant 3:178-185, 2003
15. Szklo M, Nieto FJ: Stratification and adjustment:
Multivariate analysis in epidemiology, in Epidemiology:
Beyond the Basics. Gaithersburg, MD, Aspen Publishers,
2000, p 257
16. SPSS 9.0 Advanced Models, Application Guide. Chi-
cago, IL, SPSS Inc., 1999
17. Szklo M, Nieto FJ: Defining and assessing heteroge-
ABBOTT ET AL
neity of effects: Interaction, in Epidemiology: Beyond the
Basics. Gaithersburg, MD, Aspen Publishers, 2000, p 242
18. Ojo AO, Meier-Kriesche HU, Hanson JA, et al: The
impact of simultaneous pancreas-kidney transplantation on
long-term patient survival. Transplantation 71:82-90, 2001
19. Ball AM, Gillen DL, Sherrard D, et al: Risk of hip
fracture among dialysis and renal transplant recipients. JAMA
288:3014-3018, 2002
20. Del Pizzo JJ, Jacobs SC, Bartlett ST, et al: Urological
complications of bladder-drained pancreatic allografts. Br J
Urol 81:543-547, 1998
21. Baktavatsalam R, Little DM, Connolly EM, et al:
Complications relating to the urinary tract associated with
bladder-drained pancreatic transplantation. Br J Urol 81:219-
223, 1998
22. Hosmer DW, Hosmer T, Le Cessie S, et al: A compari-
son of goodness-of-fit tests for the logistic regression model.
Stat Med 16:965-980, 1997
23. Foley RN, Guo H, Snyder JJ, et al: Septicemia in the
United States dialysis population, 1991 to 1999. J Am Soc
Nephrol 15:1038-1045, 2004
24. Stenvinkel P, Pecoits-Filho R, Lindholm B: Coronary
artery disease in end-stage renal disease: No longer a simple
plumbing problem. J Am Soc Nephrol 14:1927-1939, 2003
25. Meier-Kriesche HU, Baliga R, Kaplan B: Decreased
renal function is a strong risk factor for cardiovascular death
after renal transplantation. Transplantation 75:1291-1295,
2003
26. Abbott KC, Yuan CM, Taylor AJ, et al: Early renal
insufficiency and hospitalized heart disease after renal trans-
plantation in the era of modern immunosuppression. J Am
Soc Nephrol 14:2358-2365, 2003
27. Muller V, Becker G, Delfs M, et al: Do urinary tract
infections trigger chronic kidney transplant rejection in
man? J Urol 159:1826-1829, 1998
28. Giral M, Pascuariello G, Karam G, et al: Acute graft
pyelonephritis and long-term kidney allograft outcome. Kid-
ney Int 61:1880-1886, 2002
29. Matas AJ, Humar A, Gillingham KJ, et al: Five
preventable causes of kidney graft loss in the 1990s: A
single-center analysis. Kidney Int 62:704-714, 2002