Customize your JAMA Network experience by selecting one or more topics from the list below.
Stanford JL, Feng Z, Hamilton AS, Gilliland FD, Stephenson RA, Eley JW, Albertsen PC, Harlan LC, Potosky AL. Urinary and Sexual Function After Radical Prostatectomy for Clinically Localized Prostate CancerThe Prostate Cancer Outcomes Study. JAMA. 2000;283(3):354–360. doi:10.1001/jama.283.3.354
Context Patients with prostate cancer and their physicians need knowledge of
treatment options and their potential complications, but limited data on complications
are available in unselected population-based cohorts of patients.
Objective To measure changes in urinary and sexual function in men who have undergone
radical prostatectomy for clinically localized prostate cancer.
Design The Prostate Cancer Outcomes Study, a population-based longitudinal
cohort study with up to 24 months of follow-up.
Setting Population-based cancer registries in 6 geographic regions of the United
Participants A total of 1291 black, white, and Hispanic men aged 39 to 79 years who
were diagnosed as having primary prostate cancer between October 1, 1994,
and October 31, 1995, and who underwent radical prostatectomy within 6 months
of diagnosis for clinically localized disease.
Main Outcome Measures Distribution of and change in urinary and sexual function measures reported
by patients at baseline and 6, 12, and 24 months after diagnosis.
Results At 18 or more months following radical prostatectomy, 8.4% of men were
incontinent and 59.9% were impotent. Among men who were potent before surgery,
the proportion of men reporting impotence at 18 or more months after surgery
varied according to whether the procedure was nerve sparing (65.6% of non–nerve-sparing,
58.6% of unilateral, and 56.0% of bilateral nerve–sparing). At 18 or
more months after surgery, 41.9% reported that their sexual performance was
a moderate-to-large problem. Both sexual and urinary function varied by age
(39.0% of men aged <60 years vs 15.3%-21.7% of older men were potent at ≥18
months [P<.001]; 13.8% of men aged 75-79 years
vs 0.7%-3.6% of younger men experienced the highest level of incontinence
at ≥18 months [P = .03]), and sexual function
also varied by race (38.4% of black men reported firm erections at ≥18
months vs 25.9% of Hispanic and 21.3% of white men; P
Conclusions Our study suggests that radical prostatectomy is associated with significant
erectile dysfunction and some decline in urinary function. These results may
be particularly helpful to community-based physicians and their patients with
prostate cancer who face difficult treatment decisions.
Prostate cancer is the most frequently diagnosed solid tumor in US men.
An estimated 179,300 men will be diagnosed as having the disease in 1999,1 and in more than 70% of these patients, the disease
will be clinically localized.2 Treatment options
for men with tumors confined to the prostate who have at least a 10-year life
expectancy include radical prostatectomy, external beam radiation, brachytherapy,
or expectant management. Each of these approaches is associated with a different
spectrum of morbidity and effects on quality of life, which may be short-term
To make informed choices about treatment alternatives, patients with
prostate cancer and their physicians need accurate information to assess the
potential and pattern of complications associated with each option. Numerous
investigators have assessed urinary and sexual function 1 or more years after
radical prostatectomy, with rates of incontinence ranging from 4% to 40% and
impotence from 29% to 75%.3-12
These findings reflect the experiences of patients from selected clinical
a health maintenance organization,10 and Medicare
recipients.6,11 Differences in
patient mix, study size, and data collection methods may explain the wide
range of results.
Limited data are available to describe the outcome experiences of unselected
population-based patients. We report results from the multicenter Prostate
Cancer Outcomes Study (PCOS), which has completed longitudinal assessments
of functional status in a large community-based cohort of patients with prostate
cancer treated with radical prostatectomy for clinically localized disease.
The methods used in the PCOS have been described elsewhere.13 Briefly, African American, white, and Hispanic men
diagnosed as having primary prostate cancer between October 1, 1994, and October
31, 1995, who were residents of areas covered by 6 population-based Surveillance,
Epidemiology, and End Results cancer registries were eligible. At the time
of diagnosis, patients resided in Connecticut, New Mexico, Utah, the Atlanta,
Ga, metropolitan area, Los Angeles County, California, or King County, Washington,
which includes Seattle. All men aged 39 to 89 years were eligible in 5 registries
and those aged 60 to 89 years were eligible in King County. The study was
approved by the institutional review board of each participating institution.
A total of 11,137 eligible cases were identified, and 5672 were randomly
sampled for PCOS according to defined age and race/ethnicity strata. Of the
sampled cases, 4736 (83.5%) were contacted and invited to participate, and
3533 (62.3%) completed a 6- and/or 12-month survey. The reasons for nonresponse
were inability to locate the patient (n = 413), physician refusal (n = 380),
patient refusal (n = 1087), illness or mental incompetence of the patient
(n = 143), and other reasons (n = 116). Medical record abstracts were completed
for 3486 (98.7%) of the sampled, participating cases.
For this analysis of surgery, we selected all PCOS patients aged 39
to 79 years with histologically confirmed, clinically localized prostate cancer
who underwent radical prostatectomy as primary treatment within 6 months of
diagnosis date and who had both survey and medical records data (n = 1301).
We excluded 10 patients with missing data on key variables, leaving 1291 patients
Following physician notification, eligible sampled patients were contacted
by mail (90.2%) or telephone/in-person (8.8%) 6 months after diagnosis date
and asked to complete a self-administered questionnaire and provide consent
for access to medical records. Men who completed the 6-month survey and those
who were sampled but who did not complete a 6-month survey were mailed a 12-month
survey. At approximately 24 months following diagnosis, patients who had completed
an earlier survey were mailed another follow-up questionnaire. Of the 1291
patients in this analysis, 1042 (80.7%) completed a 6- and/or 12-month survey
and a 24-month survey.
The 6-month survey obtained information on demographics; treatment of
prostate cancer; medical history; urinary, bowel, and sexual function at baseline
(just before diagnosis) and during the past month; satisfaction with treatment;
and quality of life. Survey questions related to function were adapted from
previously used reliable and validated instruments.6,12,14
The 12-month and 24-month surveys contained similar questions that focused
on functional status during the past month. The interval from diagnosis to
survey closely approximated the interval from surgery to survey. At the 12-
and 24-month surveys, the median time since surgery was 10.5 and 22.5 months,
respectively. Patients had undergone prostatectomy at least 18 months previously
when the 24-month survey was completed.
Medical record abstracts ascertained details of the prostate cancer
diagnosis and treatment and clinical characteristics of the disease. Office
and hospital records were used to define stage of disease, tumor grade, primary
treatment, and dates of therapy.
To assess temporal changes in urinary and sexual function and potentially
confounding or modifying factors that affect these functions after radical
prostatectomy, longitudinal regression models15
implemented in SUDAAN16 were used. The Horvitz-Thompson
weight, that is, the inverse of the sampling proportion for each sampling
stratum (defined by age, race, and study center), was used to obtain unbiased
estimates of the regression parameters for the target population of patients
with prostate cancer undergoing radical prostatectomy who were eligible for
Dependent variables were urinary (level of urinary control, frequency
of incontinence, use of pads, frequency of urination, and extent of any problem
with incontinence) and sexual (interest in sex, frequency of sex, firmness
of erections, difficulty maintaining erections, and extent of any problem
with sexual function) function measured at 6, 12, and 24 months. In addition,
the first 4 items in each category were combined to yield a composite score,10 ranging from 0 to 100. A score of 100 on the sexual
scale would represent a man who reported interest in sex, sexual activity
several times a week, erections firm enough for sexual intercourse, and no
difficulty maintaining erections. A score of 40 may represent a man who was
interested in sex and was engaging in some form of sexual activity but whose
erections were not firm enough for intercourse. Independent variables used
in the analysis were survey time, baseline values of the dependent variables,
age, race, marital status, income, education, employment status, clinical
stage, and pathologic grade.
For continuous, binary, and multinomial responses, linear regression,
logistic regression, and multinomial logistic regression models, respectively,
were used. The statistical significance level for each covariate effect on
each dependent variable was examined. The interaction effects of period by
baseline measures and period by important model-identified covariates were
examined by adding these interaction terms to the model. A change in the mean
function score from baseline was tested by a paired t
test. All P values were 2-sided.
The distributions of patients with prostate cancer according to selected
demographic and clinical characteristics are provided in Table 1. A total of 56.4% of patients were younger than 65 years
at diagnosis (mean age at diagnosis, 62.9 years). Most patients were white,
married, high school or college educated, of middle income, and retired. Based
on surgical pathologic findings, most of the tumors were Gleason score 7 (31.1%)
and were of local stage (66.8%). Arthritis and hypertension were the most
prevalent medical conditions reported.
Urinary function was assessed by 5 questions about each of the 4 different
periods. The proportion of men reporting total urinary control increased after
surgery from 20.5% at 6 months to 31.9% by 24 months (Table 2). Overall, 40.2% of the patients reported occasional urinary
leaking, 6.8% frequent urinary leaking, and 1.6% no urinary control 24 months
after diagnosis (≥18 months after surgery). At the time of the 24-month
survey, 11.9% experienced incontinence more than 2 times daily, and 3.3% required
3 or more pads per day. Overall urinary function based on the composite score
decreased from 91.2 at baseline to 75.1 at 24 months (P<.001). The level of bother improved over time, with 8.7% of the
patients reporting that incontinence was a moderate-to-big problem at 24 months.
Regression models were used to examine the effects of selected covariates
on measures of urinary function. As expected, the period after surgery was
the strongest predictor of urinary function. Income was related to the mean
incontinence function score, with men who had higher incomes reporting better
function at all time points compared with lower-income groups (P = .04). Marital status was also associated with urinary function,
with 33.2% of married men reporting no incontinence at 24 months compared
with 26% of unmarried men (P = .006). Race, education,
employment status, and tumor stage and grade were not associated with urinary
Age was related to the level of urinary control, frequency of incontinence,
and bother. We evaluated measures of urinary function according to age and
period in more detail (Table 3).
Compared with younger men, those aged 75 to 79 years experienced the highest
level of incontinence at 24 months (13.8% vs 0.7%-3.6%, P = .03). The oldest age group also had more frequent incontinence
(40.8% vs 10.0%-15.9% of younger men experienced incontinence more than twice
a day, P<.001). Men younger than 60 years were
less likely to be incontinent at 24 months than older men, as measured by
total incontinence (0.7%) or frequent (>2 times per day) incontinence (10.0%).
The effects of age on urinary control (P = .03) and
frequency of incontinence (P = .008) were statistically
significant. In addition, there were statistically significant age-by-period
interactions for these 2 measures of urinary function, with younger men regaining
function sooner than older men.
The surveys also asked about treatment for strictures and incontinence.
Overall, 16.1% of the cohort reported surgery for treatment of strictures,
and 15.8% used medication to treat incontinence during the 24-month follow-up
Sexual function was assessed by several questions (Table 4). Analyses were completed for all men in the cohort and
for only those men who reported a sexual partner at the time of the baseline
or 6-month survey (80.9% of the cohort). Since the findings were similar,
only the results for the entire cohort are provided. At 24 months (≥18
months after surgery), 59.9% of men reported that erections were not firm
enough for sexual intercourse, and 44.2% were unable to have any erections
(Table 4). Of the 15.8% of men
who reported at baseline that erections were not firm enough for intercourse,
95.3% reported that they remained impotent. For comparison, of the 72.7% of
men who were potent at baseline, 72.4% reported that their erections were
not firm enough for intercourse at 18 or more months after surgery. Among
men who were potent at baseline, the proportion who were impotent 18 or more
months after surgery varied according to whether a nerve-sparing procedure
was attempted: 65.6% of non–nerve–sparing, 58.6% of unilateral
nerve–sparing, and 56.0% of bilateral nerve–sparing procedures
produced impotence (P = .001 comparing non–nerve-sparing
to bilateral nerve–sparing patients). The mean sexual function score
significantly decreased from 71.5 at baseline to 38.6 at 24 months (P<.001). Before the diagnosis of prostate cancer, 17.9%
of patients felt that sexual function was a moderate-to-big problem. On the
24-month survey, 41.9% reported that sexual functioning was a moderate-to-big
Regression models were used to examine the effects of selected covariates
on sexual function and bother. Tumor stage and grade were unrelated to sexual
outcomes. The period when the survey was completed and the level of baseline
sexual function were the strongest predictors of subsequent sexual function.
Men with better sexual function at baseline reported better sexual function
at each follow-up survey (P<.001). Age, race,
education, and having a sexual partner were also related to outcomes. The
frequency of sexual activity was strongly associated with age (P<.001), with men in the youngest (<60 years) age group reporting
more frequent sexual activity at each period compared with older men. At 24
months, a higher proportion of men younger than 65 years reported that sexual
function was a moderate-to-big problem compared with men 65 years or older
(57.1% of younger vs 48.0% of older men, P<.001).
Education also was related to sexual function (P
= .008). Men with a college or postgraduate education reported the highest
frequency of sexual activity and the lowest level of bother about sexual performance
at all periods compared with men with less education.
We performed a more detailed analysis of potency according to age, race,
and period. As shown in Table 5,
the proportion of men potent at 24 months was higher in those younger than
60 years (39.0%) compared with the older age groups (15.3%-21.7%; P<.001). The effect of age on erectile function (P = .02) and the age-by-period interaction (P<.001)
was statistically significant, indicating that the youngest men (<60 years)
had better outcomes and regained sexual function sooner than older men.
Function also varied by race, with African American men experiencing
a better outcome (Table 5). A
total of 38.4% of blacks reported firm erections at 24 months compared with
only 21.3% of whites and 25.9% of Hispanics (P =
We also examined the proportions of men who reported using aids for
erectile dysfunction. During the 24-month follow-up, these treatments were
used: vacuum suction device, 26.8%; penile injections, 21.4%; medication,
9.0%; counseling by a sex therapist or psychologist, 7.6%; and penile implant
or prosthesis, 3.7%.
Patients were asked about general effects of prostate cancer and its
treatment on other aspects of life. Small proportions of men reported experiencing
a lot of physical discomfort (3.1%), were worried about their health (5.0%),
were limited in their daily activities (3.1%), or were bothered a lot by the
prostate cancer or its treatment (5.6%) at 24 months. About half of the patients
were delighted or pleased with their surgery, and only 4% were dissatisfied.
At 18 or more months after surgery, 71.5% of the patients said that they would
make the same treatment choice, although this varied by race (blacks, 56.4%;
whites, 76.1%; and Hispanics, 61.4%). Only 7.3% reported that they would not
choose radical prostatectomy again.
These results provide the first description, to our knowledge, of outcomes
experienced by a cohort of unselected, population-based patients who have
undergone radical prostatectomy. Our findings are likely to be more representative
of the occurrence of urinary and sexual dysfunction following surgery for
localized prostate cancer in men in the community compared with earlier studies,
which were limited to single or multiple institutional-based patient series.
Overall, we found that at 18 or more months after surgery, 8.4% of patients
were incontinent and 59.9% were impotent. The proportion of men bothered by
the lack of urinary control was 8.7% at 24 months; however, 41.9% reported
that sexual function was a moderate-to-big problem. Urinary function varied
by age and sexual function by age and race. Despite significant declines in
both urinary and sexual performance levels, most men were satisfied with their
In addition to its population-based design, another strength of the
study is the large number of patients available, particularly those younger
than 65 years. These attributes provide the opportunity to adequately evaluate
outcomes in community-based patients who choose prostatectomy as primary treatment.
For this study, we collected self-reported outcomes data directly from patients
rather than relying on medical records, which may not adequately record functional
status following treatment.17,18
Further, we obtained information to assess potential confounding and modifying
factors and baseline information on functional status, which allowed us to
compare outcomes before and after diagnosis and treatment.
Our study also has several limitations. Overall, 62.3% of eligible,
sampled men participated in the survey. It is possible that nonparticipants
experienced different levels of urinary and sexual dysfunction than were reported
by participants. In comparing those who did with those who did not complete
a study survey, however, there were no major differences in the distributions
of age, race, or stage of disease. In addition, a 24-month survey was not
completed by 19.2% of those who had completed an earlier survey. To assess
possible bias, we examined the distributions of demographic factors and urinary
and sexual function at baseline and 12 months for men who did compared with
those who did not complete a 24-month survey. Men who did not complete a 24-month
survey were more likely to be older and black or Hispanic and to have less
education and a lower income. At 12 months, the nonrespondents had more incontinence
(4.5% compared with 2.4% of respondents) and less impotence (56.0% compared
with 75.5% of respondents). These differences, however, are associated with
minimal bias due to the overall high level of response on the 24-month survey.
Another potential limitation is recall bias, since baseline (prediagnosis)
function was assessed on the 6-month survey. We conducted a reliability study
on 133 patients who completed a baseline survey immediately after diagnosis
and a 6-month survey. Recall of baseline function at 6 months was identical
to baseline function reported immediately after diagnosis for 69% or more
of the men on each item. However, men overestimated baseline function at 6
months for 2 items: frequency of incontinence and erections firm enough for
intercourse. To the extent that recall bias exists in our data, the mean change
scores for incontinence and sexual function may be overestimated. Such a bias
would not affect results for other individual survey items or responses about
current function on the follow-up surveys.
Frequent urinary leakage or total incontinence was reported by 8.4%
of our cohort at 24 months (≥18 months after surgery), with 11.9% of men
experiencing incontinence more than twice a day and 3.3% using 3 or more pads
each day. These results are similar to those reported from several clinic-
and hospital-based patient series that found that 4% to 12% of patients who
had undergone prostatectomy were incontinent a year or more after treatment.3-5,7,9,12
However, our estimates of urinary dysfunction are lower than the 31%11 and 40%10 levels of
incontinence reported among older men. We found that age was a significant
predictor of urinary function, with elderly men experiencing the most incontinence.
The 3.3% of men using pads for incontinence at 24 months in our cohort
is similar to the 4% using 2 or more pads per day reported by Murphy et al,7 based on a large series of patients from multiple
institutions. In contrast, Talcott et al12
found that 35% of men were using pads 12 months after prostatectomy, and Fowler
et al11 noted that 31% of Medicare patients
surveyed 2 to 4 years after surgery reported use of pads and/or clamps. Talcott
et al12 suggested that 35% is likely to be
an overestimate, since some men wear pads as a precaution. In the study by
Talcott et al,12 incontinence 12 months after
surgery was reported by 9% of men younger than 65 years and 15% of those 65
years or older at diagnosis. Our estimate of incontinence also varied by age.
At the time of the 24-month survey, 6.6% of younger men (<65 years) were
incontinent compared with 10.7% of older (≥65 years) men.
Although the mean incontinence function score was significantly lower
at 24 months compared with baseline, only 8.7% of our cohort reported that
incontinence was a moderate-to-big problem 24 months after diagnosis. This
estimate is similar to the 6% of patients bothered by incontinence at 12 months
after treatment in the patient series reported by Braslis et al9
but lower than the 34% reported by Jonler et al.8
The frequency of impotence in our cohort (59.9%), as measured by the
inability to have erections firm enough for sexual intercourse at 18 or more
months after prostatectomy, is higher than the estimated 32% to 53% and 39%
of men experiencing impotence reported by Catalona et al5
and Steiner et al,3 respectively. However,
both of these studies were based on patient series from single institutions.
Our cohort also reported a higher level of erectile dysfunction than the 52%
noted in health maintenance organization members10
and the 46% reported for Medicare patients.11
Among men who were potent at baseline, the proportion who were impotent at
18 or more months after radical prostatectomy differed somewhat according
to whether the surgery was nerve sparing (65.6% non–nerve-sparing, 58.6%
unilateral, and 56.0% bilateral nerve-sparing patients were impotent). These
results are similar to some previous studies that showed that potency differs
by type of procedure.5,19 Sexual
dysfunction also varied by age and race. In men younger than 65 years, 55.1%
reported impotence compared with 66.1% of men 65 years and older. In addition,
as of the 24-month survey, 37.6% of younger men (<65 years) compared with
52.6% of older men (≥65 years) reported that they were unable to have any
erections. The frequency of firm erections at 24 months was higher in African
American men (38.4%) compared with whites (21.3%) or Hispanics (25.9%). Our
study is, to our knowledge, the first to examine functional status in minority
The mean sexual function score in our cohort declined from 71.5 at baseline
to 38.6 by 18 or more months after surgery (P<.001).
These results did not differ notably when examining all men in the cohort
compared with those who reported a sexual partner at baseline or 6 months.
An assessment of bother showed that at 24 months, 41.9% of men felt that their
sexual function was a moderate-to-big problem. Younger men (<65 years)
were more likely to report that sexual function was a moderate-to-big problem
compared with men 65 years and older (45.4% vs 37.3%), although younger men
reported better sexual function at all time points compared with their older
Despite the level of urinary incontinence and sexual dysfunction reported
on the 24-month survey, most men (75.5%) were satisfied or pleased with their
treatment, and most (71.5%) would choose radical prostatectomy again. These
results on satisfaction are similar to other reports.8,11
In earlier studies, 88% to 92% of men undergoing prostatectomy reported that
they would select surgery again.8,9,11,20
Our lower estimate may reflect differences in study design, size, and patient
In summary, our results provide new information on urinary and sexual
function at multiple points after radical prostatectomy in a large population-based
cohort of patients with clinically localized disease. These estimates of incontinence
and sexual dysfunction measured at 18 or more months after surgery should
be particularly helpful to community-based patients faced with difficult treatment
options. Since follow-up of this population-based cohort is ongoing, we hope
to provide more information on functional status in the future. The ability
to obtain long-term follow-up information on this large cohort should provide
new insights into the temporal changes in function that can be used to predict
outcomes in similar patients undergoing radical prostatectomy.