934 1030
Postdiagnosis models were further adjusted for the first course of
treatment, education, family history of prostate cancer, any history of
prediagnosis PSA testing not leading to prostate cancer diagnosis,
postdiagnosis history of CVD, other cancer, postdiagnosis smoking
status, BMI, and prediagnosis red and processed meat consumption
(postdiagnosis intake was not available for cases diagnosed after 2003)
as shown in Supplementary Table 1. When walking and other
recreational PA were examined separately, they were mutually adjusted.
For trend analysis, median values of each PA category were modeled as a
continuous variable.
Stratified analyses were conducted to examine the associations
with all outcomes within two tumor risk categories based on National
Comprehensive Cancer Network guidelines
[23] ,reflecting the
likelihood of death from prostate cancer
[24] .The high-risk group
comprised patients diagnosed with Gleason score 8–10 or T3–T4 or N1
cancer, and the lower-risk group comprised patients diagnosed with
Gleason score 2–7 and T1–T2 cancer. The
p
values for
[9_TD$DIFF]
interaction were
calculated by the likelihood ratio test comparing a model
[2_TD$DIFF]
with
[16_TD$DIFF]
interaction
[17_TD$DIFF]
terms
[18_TD$DIFF]
between
[19_TD$DIFF]
recreational
[20_TD$DIFF]
PA
[21_TD$DIFF]
and
[22_TD$DIFF]
tumor
[23_TD$DIFF]
risk
[24_TD$DIFF]
category
[25_TD$DIFF]
and a
[26_TD$DIFF]
model
[27_TD$DIFF]
without
[25] .The proportional hazards assumption was
examined by testing the interaction terms of recreational PA and
[28_TD$DIFF]
continuous survival time using the likelihood ratio test. Interaction
effects on PCSM were also examined using the likelihood ratio test
between recreational PA and age at diagnosis, BMI, smoking, diagnosis
year, and follow-up time. All analyses were conducted using SAS 9.3
(SAS Institute, Cary, NC, USA).
3.
Results
Most men were elderly and white
( Table 1). The median age
at prostate cancer diagnosis was 71 yr (interquartile range:
67–75 yr). Through 2012, there were 2743 deaths (454 due
to prostate cancer and 754 due to CVD) among 7328 men
included in the prediagnosis analysis and 1685 deaths
(261 due to prostate cancer and 464 due to CVD) among
5319 men included in postdiagnosis analysis
( Fig. 1 ). In
prediagnosis analysis, median time from diagnosis to death
date or end of follow-up was 7.8 yr for men who died from
prostate cancer and 10.3 yr for those who did not. In
postdiagnosis analysis, median time from 4 yr after
postdiagnosis survey completion to death date or end of
follow-up was 3.8 yr for men who died from prostate cancer
and 6.5 yr for those who did not. Compared with men who
were inactive, those who reported higher levels of
recreational PA before diagnosis were more likely to have
been diagnosed with lower-risk tumors (T1–T2 and Gleason
score 2–7); to have a history of PSA testing; to be leaner,
nonsmokers, and current multivitamin supplement users;
and to consume more fish but less red and processed meat
( Table 1). Compared with men excluded from postdiagnosis
analysis (
n
= 4756), those included had lower stage and
Table 1 (
Continued
)
Prediagnosis recreational physical activity (MET-h/wk)
<
3.5
(
n
= 865)
N
(%)
3.5–8.75
(
n
= 2312)
N
(%)
8.75–
<
17.5
(
n
= 1614)
N
(%)
17.5
(
n
= 2537)
N
(%)
Prediagnostic smoking status
Never
262 (30.3)
810 (35.1)
561 (34.7)
958 (37.7)
Current
88 (10.4)
152 (6.3)
84 (5.3)
101 (4.1)
Former
512 (59)
1343 (58.3)
961 (59.5)
1470 (58)
Unknown
3 (0.4)
7 (0.3)
8 (0.5)
8 (0.3)
Prediagnostic multivitamin use
No
511 (60.2)
1301 (55.1)
851 (53.3)
1257 (50.3)
Yes
280 (31.2)
869 (38.3)
649 (39.8)
1132 (44.1)
Missing
74 (8.6)
142 (6.6)
114 (6.9)
148 (5.6)
Prediagnostic red and processed meat intake
Q1
141 (16.3)
448 (19.3)
359 (22.3)
669 (26.4)
Q2
155 (18.1)
542 (23.2)
365 (22.7)
571 (22.7)
Q3
167 (19.1)
531 (22.8)
364 (22.6)
548 (21.6)
Q4
274 (32)
528 (22.9)
345 (21.4)
486 (19.1)
Missing
128 (14.5)
263 (11.9)
181 (11.1)
263 (10.3)
Prediagnostic fish intake
Q1
251 (29.2)
574 (24.7)
363 (22.3)
552 (21.7)
Q2
175 (20.4)
477 (20.2)
311 (19.5)
437 (17.6)
Q3
191 (22.1)
554 (24.3)
409 (25.2)
691 (27)
Q4
120 (13.9)
444 (19)
350 (21.8)
594 (23.5)
Missing
128 (14.5)
263 (11.9)
181 (11.1)
263 (10.3)
Prediagnostic leisure time sitting (h/wk)
<
3
341 (39.6)
950 (41.4)
666 (41.3)
990 (39.3)
3–
<
6
303 (35.3)
952 (40.3)
636 (39.5)
990 (39.2)
6+
157 (17.8)
328 (14.5)
262 (16.2)
475 (18.4)
Missing
64 (7.3)
82 (3.8)
50 (3)
82 (3.2)
CPS = Cancer Prevention Study; PSA = prostate-specific antigen; MET = metabolic equivalent task.
a
Standardized on calendar period of diagnosis unless otherwise indicated.
b
Not standardized on calendar period of diagnosis.
c
Based on self-reports of PSA test made on biennial follow-up questionnaires completed before diagnosis of prostate cancer. For the 1997 survey (when PSA
testing was first asked), it was the testing that occurred at least 180 d before diagnosis.
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 9 3 1 – 9 3 9
934