Fax: (617) 432-2435
The association of treatment-related symptoms with quality-of-life outcomes for localized prostate carcinoma patients
Article first published online: 31 JAN 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 3, pages 862–871, 1 February 2002
How to Cite
Bacon, C. G., Giovannucci, E., Testa, M., Glass, T. A. and Kawachi, I. (2002), The association of treatment-related symptoms with quality-of-life outcomes for localized prostate carcinoma patients. Cancer, 94: 862–871. doi: 10.1002/cncr.10248
- Issue published online: 31 JAN 2002
- Article first published online: 31 JAN 2002
- Manuscript Accepted: 28 OCT 2001
- Manuscript Revised: 10 OCT 2001
- Manuscript Received: 10 JAN 2001
- American Cancer Society. Grant Number: RPG98101-01-GR
- National Institutes of Health. Grant Numbers: CA55075, HL35464
- quality of life;
- prostatic neoplasm;
- urinary incontinence;
- erectile dysfunction;
- bowel symptoms
Most studies of treatment outcomes in men with localized prostate carcinoma have emphasized sexual, urinary, and bowel symptoms with the assumption that they have an impact on quality of life. However, very few studies have directly examined and compared the impact of these symptoms on overall and cancer specific quality of life.
The authors examined 783 incident cases of localized prostate carcinoma, diagnosed from 1993 to 1998, and 1928 age-matched healthy controls from the Health Professionals Follow-Up Study cohort. Information on frequency of ejaculation and urinary symptoms were collected before cancer diagnosis. After cancer diagnosis, the authors mailed a questionnaire including the Medical Outcomes Study Short Form–36 Health Status Survey (SF-36), the Cancer Rehabilitation Evaluation System–Short Form (CARES-SF), and the University of California at Los Angeles Prostate Cancer Index in 1998.
Cases had slightly lower scores on most of the SF-36 scales and reported much more bother from sexual, urinary, and bowel symptoms compared with healthy controls. Among prostate carcinoma patients, bowel symptoms had the greatest negative impact on quality of life, followed by sexual and urinary symptoms. As expected, treatment-related symptoms were associated with the physical domains of quality of life, but psychosocial domains were just as strongly affected.
Patients and health care providers need to consider the potential mental quality-of-life impacts associated with prostate carcinoma treatment symptoms when making treatment decisions. Even after patients have completed cancer treatment, significant health impairments may remain. Health care providers should continue to address the mental and physical well-being of prostate carcinoma patients in follow-up care. Cancer 2002;94:862–71. © 2002 American Cancer Society.
Localized prostate carcinoma treatments have been shown to deleteriously affect sexual, urinary, and bowel function for many men.1–8 Because the prognosis associated with localized prostate carcinoma is excellent (5-year relative survival rate nearly 100%9), quality-of-life issues become a key factor in decision making regarding the choice of treatment.
Most outcomes research among men receiving prostatectomy has focused on sexual and urinary function.3, 5, 6, 8, 10–14 These symptoms, as well as bowel function, are also relevant for other localized prostate carcinoma treatment options, including external beam radiation, brachytherapy, hormone therapy, and watchful waiting. Symptoms associated with therapy for localized prostate carcinoma vary widely depending on factors such as patient age, treatment type, tumor stage, and length of time since treatment. For instance, the reported rate of erectile dysfunction in men who have had prostatectomy ranges from 44% to 80%.3, 5–7, 13, 15 Reported levels of incontinence range from 30% to 70% 1 year after diagnosis and treatment.3–7, 11, 15 Bowel dysfunction has been noted to occur in approximately 30–45% of patients treated with external beam radiation,4, 7, 16 and to a lesser degree with other types of treatment.4, 6, 7, 15
Most studies of treatment outcomes in men with prostate carcinoma have emphasized sexual, urinary, and bowel symptoms with the assumption that they have an impact on quality of life. However, very few studies have directly examined and compared the impact of these symptoms on overall and cancer specific quality of life. In one of the few studies that examined this issue, Lim and colleagues17 found that problems with incontinence, sexual function, and bowel function were associated with depression, tension, and fatigue. More recently, two studies found that incontinence and impotence were associated with reduced overall quality of life for prostatectomy patients.3, 11
In the study presented here, we examined the impact of common prostate carcinoma treatment-associated symptoms on generic health-related quality of life, as well as cancer specific quality of life. In particular, we addressed: 1) what types of treatment-related symptoms (e.g., erectile dysfunction, urinary incontinence, and bowel symptoms) are associated with the largest adverse effects on quality of life; 2) which domains of quality of life are most affected by symptomatology; and 3) whether posttreatment quality of life was modified by prediagnosis levels of sexual and urinary function.
Our study population was drawn from the Health Professionals Follow-Up Study (HPFS), a cohort established in 1986 with the goal of examining risk factors for heart disease and cancer. The study population of 51,529 men in the U.S., ages 40–75 years old in 1986, includes dentists (58%), veterinarians (20%), pharmacists (8%), optometrists (7%), osteopaths (4%), and podiatrists (3%). Since the original questionnaire was mailed in 1986, follow-up questionnaires have been mailed every 2 years with a response rate of 94%. Information collected on the biennial surveys includes health behaviors and medical history of major illnesses.
Prostate Carcinoma Ascertainment
For the purpose of this study, we included incident cases of localized prostate carcinoma confirmed via medical record occurring between January 1993 through December 1998. Of the 1489 men in the HPFS cohort with prostate carcinoma diagnosed during the study period, 1201 were confirmed localized prostate carcinoma cases via independent review of medical records and pathology reports obtained with the participants' permission.
For each newly diagnosed prostate carcinoma case, we surveyed two healthy control subjects from the HPFS cohort. Controls were randomly selected from an age-matched (in 5-year intervals) pool of HPFS participants who were free of cancer, cardiovascular disease, diabetes, stroke, renal failure, Parkinson's disease, muscular dystrophy, or amyotrophic lateral sclerosis.
Supplemental questionnaires were mailed in 1998 to 1201 cases and 2412 controls, including 2 additional mailings to nonrespondents. The questionnaires asked participants about their generic and cancer specific quality of life, cancer treatment and recurrence history, and sexual, urinary, and bowel symptoms. After 3 mailings, responses were received from 92% of cases and 93% of controls, including notification of death. Complete data were obtained for 783 localized prostate carcinoma cases and 1928 healthy controls. The primary reasons for exclusion included incomplete data from primary predictor or outcome variables (18%), refusal to participate (1%), and death (0.4%).
In addition to data gathered from the supplemental questionnaire, we used information gathered on the regular HPFS questionnaires mailed to the entire cohort. These data include disease history, health behaviors, and demographic factors. More detail on the HPFS questionnaires and study design is reported elsewhere.18, 19
The patients included in this study had apparent organ-confined prostate carcinoma at the time of diagnosis. Among patients for whom information on cancer grade was available in the medical record (n = 674), greater than half (57%) had medium Gleason scores (Gleason score 5–6). Low-grade cancers (Gleason score 2–4) made up 11% of the cases, and the remaining 31% were high-grade cancers (Gleason score 7–10). According to medical chart review, initial prostate carcinoma treatments included prostatectomy (51%), external beam radiation (26%), brachytherapy (8%), hormone therapy (4%), watchful waiting (3%), and other treatment (7%) (self-reported treatment used if medical chart incomplete or unavailable).
Assessment of Sexual, Urinary, and Bowel Symptoms
We administered the UCLA Prostate Cancer Index to measure the level of sexual, urinary, and bowel symptoms related to prostate carcinoma therapy.1 The sexual function scale includes eight items related to sexual desire, frequency, and ability to perform sexual intercourse. The urinary function scale consists of five items related to control and leakage. The bowel function scale includes four items assessing urgency, consistency, and pain. Each of the scales range from 0 to 100, with higher scores denoting higher levels of functioning.
The reliability of the University of California at Los Angeles instrument has been tested in two different samples of prostate carcinoma patients.1, 20 The larger of the two studies (n = 2382) found that the scales had high internal consistency, with Cronbach α coefficients of 0.87 for sexual function, 0.88 for urinary function, and 0.78 for bowel function. Test–retest reliability (14 days or less) in a subgroup of 348 prostate carcinoma patients found correlation coefficients for the scales ranging from 0.74 for bowel function to 0.93 for sexual function.
The American Urological Association (AUA) Symptom Index21 was measured in 1994 in the entire cohort, before the diagnosis of prostate carcinoma for participants in this study, using a slightly modified format to fit within the constraints of the HPFS questionnaire. The AUA Symptom Index is a seven-item scale that measures urinary symptoms, including frequency, nocturia, weak urinary stream, hesitancy, intermittence, incomplete emptying, and urgency. The AUA Symptom Index ranges from 0 (no symptoms) to 35 (maximum symptoms). Lower urinary tract symptoms were categorized as none to mild (AUA scores from 0 through 7), low to moderate (8–14), high to moderate (15–19), or severe (20–35). High internal consistency (Cronbach α = 0.86) and test–retest reliability (r = 0.92) have been reported for this scale.21, 22
As a proxy for the level of sexual activity before prostate carcinoma diagnosis, we asked about frequency of ejaculation on the 1992 survey using the following question: “On average, how many ejaculations did you have per month during the past year?” with response options: none; 1–3; 4–7; 8–12; 13–20; and greater than 20 per month.
Assessment of Quality-of-Life Outcomes
General health-related quality of life
On the 1998 supplemental questionnaire, we included the Medical Outcomes Study Short Form–36 Health Status Survey (SF-36) as a measure of generic health-related quality of life.23 Eight domains of quality of life are measured by the SF-36: physical function (10 items), role limitation due to physical problems (4 items), bodily pain (2 items), general health perceptions (5 items), vitality (4 items), social function (2 items), role limitation due to emotional problems (3 items), and mental health (5 items). Each of the SF-36 scales was scored separately with a range of 0 (worst health) to 100 (best health).
The SF-36 is a widely used instrument that has been extensively tested for reliability and validity.20, 24–28 The SF-36 has been shown to discriminate between types and levels of disease,29, 30 as well as between persons with chronic medical conditions alone versus those with both physical and psychologic conditions.31
From the original 36 items, 2 SF-36 summary scales also were calculated. The physical health component score (PCS) and the mental health component score (MCS) are standardized to the U.S. general population, with a mean score of 50 and standard deviation equal to 10.32
Scoring of the SF-36 scales was conducted by following the procedures described in the SF-36 manual.23 When less than 50% of the items for a particular SF-36 scale were missing, we imputed the overall score by substituting the person specific mean scores for the completed items on that particular scale. When greater than half of the items for a scale were missing, that scale was recorded as missing for that individual.
Cancer specific quality of life
The Cancer Rehabilitation Evaluation System–Short Form (CARES-SF) was developed for use with cancer patients.33 Designed specifically for research, the short version retains the major content domains covered by the original 139-item instrument related to the daily problems and rehabilitation needs of cancer patients.34 We included this instrument only in the supplemental questionnaire mailed to patients with prostate carcinoma. The CARES-SF includes 59 problem statements, each of which is rated on a 5-point Likert scale ranging from “not at all” (no problem) to “very much” (severe problem).
Items of the CARES-SF were combined into a global summary score. We also analyzed the following four specific subscales: 1) physical changes and disruption of daily activity caused by the disease or treatment (10 items); 2) psychosocial issues, communication, and relationship problems (17 items); 3) medical interaction problems (4 items); and 4) marital problems (6 items). Both the global summary score and individual subscale scores range from 0 (best health) to 100 (worst health).
In a sample of lung, prostate, and colorectal carcinoma patients, the reliability of the instrument ranged from acceptable to good. Internal consistency measured by Cronbach α ranged from 0.67 for the medical interaction scale to 0.85 for the physical and psychosocial scales.34 The test–retest correlations for the global score was 0.92, ranging from 0.69 (sexual) to 0.87 (physical and psychosocial) for the individual scales.
Validity of the CARES-SF has not been tested as extensively as the validity of the long form. However, the CARES-SF correlated in the expected direction when compared with other commonly used instruments such as the Karnofsky performance status scale, the Symptom Checklist–90 (SCL-90), and the Functional Living Index–Cancer.33, 34
In addition to asking about the presence of treatment-related symptoms, we inquired about the level of bother caused by these symptoms. The bother component of the UCLA Prostate Cancer Index measures the amount of bother related to urinary, sexual, and bowel symptoms specifically among prostate carcinoma patients.1 Each of the scales ranges from 0 to 100, with higher scores denoting less bother.
The UCLA Prostate Cancer Index has not been extensively tested for reliability and validity compared with the SF-36 and CARES-SF. One study by the developers of the instrument reported that the test–retest reliability of the sexual, urinary, and bowel bother items after 4 weeks were 0.70, 0.66, and 0.77, respectively.35 Both Lubeck20 and Litwin35 found that the disease specific bother measures were not highly correlated with SF-36 scales, suggesting that the bother component of the UCLA Prostate Cancer Index provides additional information on aspects of quality of life not addressed in the more general measure (SF-36).
Ordinary least squares regression was used to model the impact of sexual, urinary, and bowel symptoms on SF-36, CARES-SF, and UCLA Prostate Cancer Index bother scores. Because of ceiling effects for SF-36 role function–physical (67%), role function–emotional (78%), and social function scales (69%), we used logistic regression models. For these scales, responses were dichotomized into those who reported no limitations (scale score, 100) versus those who reported any limitations (scale score, <100). Standardized regression coefficients are reported for the impact of sexual, urinary, and bowel symptoms on quality of life. The standard deviations of sexual, urinary, and bowel function for prostate carcinoma cases were 25.6, 20.7, and 16.3, respectively.
Covariates in the multivariable regression models were measured in 1998 or the most recent data available from the overall HPFS cohort questionnaires. We controlled for the following covariates: age in 1998, marital status (married/partner vs. single/widowed/divorced), waist circumference in inches, physical activity, smoking status (current, former, never), alcohol intake, and presence of comorbid conditions (heart disease, hypertension, hypercholesterolemia, diabetes mellitus, stroke, arthritis, and nonprostate carcinomas). Physical activity was measured using a validated questionnaire36 assessing the type and duration of eight common leisure activities. Activity scores are computed in metabolic-equivalent hours per week and included in the regression models in quintiles. Alcohol intake was derived from self-reported consumption converted into grams of ethanol per day, and categorized as 0, 0.1–4.9, 5.0–14.9, 15.0–29.9, and greater than or equal to 30 g. Because the interval since date of cancer diagnosis was not a statistically significant predictor of health-related quality of life in any of the multiple regression models, we did not include this variable in the final analysis.
For our analysis of the impact of posttreatment sexual function on quality-of-life outcomes, we stratified according to the level of self-reported frequency of ejaculation per month in 1992 (i.e., before prostate carcinoma diagnosis and treatment). The strata were 0 times, 1–3 times, 4–7 times, and 8 or more times per month.
Similarly, for analyses examining the impact of urinary symptoms on quality of life, we stratified according to the precancer levels of urinary symptoms as measured by the 1994 AUA Symptom Index. Because of small sample sizes at more severe levels of urinary symptoms before prostate carcinoma, categories were collapsed into two groups—men with none to mild urinary symptoms (AUA score 0–7) and men with moderate to severe symptoms (AUA score 8–35).
Characteristics of the Study Population
The distribution of demographic and health characteristics for localized prostate carcinoma patients compared with age-matched controls are shown in Table 1. Cases and controls were nearly identical in terms of marital status. Because the HPFS cohort members are all health professionals, they have similar levels of educational attainment. Cases and controls also had similar health behaviors that might have an impact on quality-of-life outcomes such as smoking status, alcohol consumption, waist circumference, and physical activity. Because of the sampling strategy, control subjects had fewer comorbid conditions compared with prostate carcinoma patients, especially heart disease (16.9% for cases vs. 0.6% for controls), hypertension (40.7% vs. 31.3%), and hypercholesterolemia (50.3% vs. 40.4%). Before development of prostate carcinoma, cases had a similar distribution of ejaculation frequency compared with the control group (6.3% of cases vs. 5.6% of controls with no ejaculations per month in 1992, and 22.3% of cases vs. 24.5% of controls with ≥8 ejaculations per month).
|Characteristic||Cases (n = 783)||Controls (n = 1928)|
|Age (mean SD yrs)||71.0||71.3|
|Waist circumference in 1996 (mean [SD] inches)||38.6||38.4|
|Physical activity (mean [SD] METS/wk)||33.5||35.4|
|Current smoker (%)||4.0%||4.9%|
|Alcohol intake (g/day)|
|30.0 or more||10.1%||10.6%|
|1992 frequency of ejaculations (times/mo)|
|8 or more||22.3%||24.5%|
|1994 American Urologic Association Symptom Index score|
|None to mild (0–7)||55.9%||55.8%|
After prostate carcinoma diagnosis and treatment, the distribution of sexual and urinary symptoms were quite different between cases and controls. As shown in Table 2, mean levels of sexual function as measured by the UCLA Prostate Cancer Index were much lower for cases than for healthy controls (29.4 vs. 56.1, P < 0.0001). Mean levels of urinary and bowel function were also lower for cases versus controls (81.7 vs. 92.7, P < 0.0001; 84.3 vs. 89.2, P < 0.0001, respectively).
|UCLA Prostate Cancer Index scales||Cases (n = 783)||Controls (n = 1928)|
|Mean (standard deviation)||29.4 (25.6)||56.1 (28.2)|
|Median (quartile 1 to quartile 3)||22.9 (7.3–48.9)||62.5 (33.4–78.1)|
|Mean (standard deviation)||81.7 (20.7)||92.7 (11.9)|
|Median (quartile 1 to quartile 3)||86.8 (68.4–100)||100 (86.8–100)|
|Mean (standard deviation)||84.3 (16.3)||89.2 (12.1)|
|Median (quartile 1 to quartile 3)||88.8 (77.0–93.8)||93.8 (85.5–100)|
As shown in Table 3, cancer free controls had slightly higher scores for 7 of 10 SF-36 scales. Role function related to physical problems and mental health were not significantly different between cases and controls. All other differences in SF-36 scores were significant (P < 0.004). Physical function and bodily pain domains of quality of life were higher (P < 0.0002) in men with localized prostate carcinoma compared with men free of major disease. As expected, cases reported significantly more bother due to sexual, urinary, and bowel symptoms compared with controls. Cancer specific quality of life was measured only among the cases, who had the worst (highest scores) quality of life in the marital interaction domain and the best (lowest scores) quality of life in the medical interaction domain. All analyses of the marital interaction scale include only men who were married or had a partner at the time of assessment.
|Outcome variable||Case mean (SD)||Control mean (SD)|
|SF-36 health-related quality-of-life scales|
|Physical function||87.0 (18.7)||85.9 (19.2)|
|Role function–physical||80.5 (33.0)||83.0 (30.8)|
|Bodily pain||82.2 (19.7)||77.4 (19.1)|
|General health||76.8 (18.3)||80.7 (15.9)|
|Vitality||67.8 (19.4)||70.8 (16.3)|
|Social function||90.2 (17.6)||92.6 (15.6)|
|Role function–emotional||87.0 (27.6)||92.4 (21.4)|
|Mental health||82.8 (14.2)||84.3 (11.9)|
|Physical summary scale||50.6 (8.5)||50.1 (8.2)|
|Mental summary scale||53.9 (8.0)||55.8 (6.5)|
|UCLA bother scales|
|Sexual bother||49.1 (38.6)||70.1 (33.9)|
|Urinary bother||81.4 (24.5)||87.3 (20.2)|
|Bowel bother||82.8 (24.8)||90.5 (19.0)|
|CARES-SF cancer specific quality-of-life scales|
|Medical interaction||4.5 (10.1)|
|Marital interaction||10.0 (12.7)|
|CARES summary||6.9 (5.4)|
Quality of Life Associated with Sexual Symptoms
Sexual symptoms were strongly associated with all quality-of-life measures except medical interaction. Reductions in quality of life were particularly evident for men with some sexual function before cancer diagnosis.
Table 4 presents the results of adjusted regression models showing the impact on SF-36 and UCLA Bother scales of a one standard deviation difference in sexual, urinary, or bowel symptoms among prostate carcinoma patients. For instance, a one standard deviation difference in sexual symptoms was associated with changes on the SF-36 scales ranging from 1.9 on the bodily pain scale (P < 0.05) to 3.9 on the vitality scale (P < 0.0001). Differences of one standard deviation in sexual symptoms were associated with a difference of more than 11 points in sexual bother (P < 0.0001). However, as shown in Table 4, sexual symptoms and the other potential confounders included in the model together only explained 12% of the variance in sexual bother. Higher sexual function also was associated with significantly lower odds of limitations in social and role functioning.
|QOL outcome||UCLA symptom scales in standard deviation units|
|Sexual function||Urinary function||Bowel function|
|Beta||SE||P value||r2||Beta||SE||P value||r2||Beta||SE||P value||r2|
|UCLA bother scales|
|Physical summary scale||1.2||0.29||<0.0001||0.26||0.8||0.27||<0.01||0.25||1.8||0.27||<0.0001||0.29|
|Mental summary scale||1.4||0.30||<0.0001||0.08||1.4||0.28||<0.0001||0.08||2.2||0.28||<0.0001||0.12|
|OR||95% CI||P value||OR||95% CI||P value||OR||95% CI||P value|
Sexual symptoms were strongly associated with SF-36 summary measures (1.2 for physical and 1.4 for mental summary scales). However, the amount of variance explained for the physical summary scale (r2 = 0.22) was much higher than for the mental summary scale (r2 = 0.08).
Sexual symptoms also were significantly related to cancer specific quality of life as assessed by the CARES-SF (Fig. 1). Because the CARES-SF is scored in the opposite direction compared with the SF-36 and the UCLA scales, a negative coefficient should be interpreted as indicating a higher level of function. Thus, sexual symptoms among prostate carcinoma patients had the strongest impact on the marital interaction scale of the CARES-SF (β = −2.4, P < 0.0001) after adjustment for potential confounders. The model of marital interaction was restricted to men who were married or had a partner. A one standard deviation difference in sexual symptoms was associated with a one- to two-point difference in CARES physical, psychosocial, and summary cancer rehabilitation scores (P < 0.0001 for all scales). Only the medical interaction scale was unrelated to sexual symptoms.
To more fully understand the quality-of-life impact of postcancer sexual function, we conducted an analysis stratified by precancer frequency of ejaculation (a proxy for precancer level of sexual activity). We found that for men who reported no sexual activity (zero ejaculations/month) before cancer diagnosis, there was no significant relation between postcancer sexual symptoms and quality-of-life outcomes for any of the SF-36 or CARES-SF scales. However, for men who reported any sexual activity before cancer diagnosis, there was a significant relation between postcancer sexual function and quality of life. The largest effects on the SF-36 scales were observed in the domains of physical function (age-adjusted β = 6.9, P < 0.0001, for men with 1–3 ejaculations per month; β = 3.1, P < 0.003, for men with 4–7 ejaculations per month), vitality (age-adjusted β = 5.6, P < 0.0003, for men with 1–3 ejaculations per month; β = 3.8, P < 0.002, for men with 4–7 ejaculations per month; and β = 4.1, P < 0.002, for men with ≥8 ejaculations per month), and mental health (age-adjusted β = 2.7, P < 0.01, for men with 1–3 ejaculations per month; β = 3.0, P < 0.001, for men with 4–7 ejaculations per month; and β = 3.1, P < 0.004, for men with ≥8 ejaculations per month). Similarly, in cancer specific domains, sexual symptoms were significantly more strongly associated with quality of life in men with any prior level of sexual activity.
Quality of Life Associated with Urinary Symptoms
Of the three types of problems assessed, urinary symptoms were the least predictive of quality of life after cancer diagnosis. Urinary function was significantly associated with all of the quality-of-life scales included in the study, but not to the same extent as sexual and bowel symptoms for most outcomes.
As shown in Table 4, a one standard deviation difference in urinary function was associated with a two- to three-point difference in physical function, general health, vitality, and mental health scales of the SF-36. The impact of urinary symptoms on the summary scales of the SF-36 was very similar to that of sexual function for the mental summary scale (β = 1.4, P < 0.0001) but weaker for the physical summary scale (β = 0.8, P < 0.01). Not surprisingly, the degree of urinary symptoms was also strongly related to the amount urinary bother (β = 16.6, P < 0.0001).
For the cancer specific outcomes measured by the CARES-SF, the strongest effects of urinary function were on the marital interaction (β = −2.7, P < 0.0001) and psychosocial scales (β = −1.9, P < 0.0001 for 1 standard deviation difference in urinary function). Statistically significant, though smaller, effects of urinary symptoms were observed for the more physical domains of the CARES-SF, including physical aspects of cancer rehabilitation and medical interaction (see darkest bars in Fig. 1). As previously indicated, a negative coefficient in the CARES-SF regression models indicate better health outcomes.
We sought to determine whether there was an interaction between the impact of urinary symptoms on quality of life, depending on urinary function before prostate carcinoma diagnosis. Because the AUA Urinary Symptoms Index was administered to the entire HPFS cohort in 1994, we were able to conduct a stratified analysis according to precancer levels of urinary symptoms. Using regression models adjusted for age only, we stratified men into two groups—those with none to mild urinary symptoms before prostate carcinoma (AUA score, 0–7) and those with moderate to severe urinary symptoms before prostate carcinoma (AUA score, 8–35). In the men with moderate to severe symptoms before cancer diagnosis, the SF-36 physical function scale and the CARES-SF summary measure were the only quality-of-life outcomes associated with urinary function after cancer diagnosis (β = 3.1, P < 0.03 and β = −1.0, P < 0.01, respectively for a 1 standard deviation difference in urinary function). However, in the men with none to mild urinary symptoms before prostate carcinoma, postcancer urinary function was significantly associated (P < 0.01) with general health perceptions (β = 2.7), vitality (β = 2.9), mental health (β = 3.5), and the mental summary scale (β = 1.6). Urinary function also was significantly associated (P < 0.01) with psychosocial (β = −2.1), marital interaction (β = −2.7), and the global summary scale (β = −1.5) of the CARES-SF in these men.
Quality of Life Associated with Bowel Symptoms
The impact of compromised bowel function (rectal urgency, loose or liquid stools, crampy pain, and bowel movement distress) on quality of life was generally more severe compared with sexual and urinary problems. For the SF-36 scales, a 1 standard deviation difference in bowel function was associated with effects ranging from a 3.1-point difference in physical function (P < 0.0001) to 5.0-point differences in vitality and general health perceptions (P < 0.0001). The odds of having reduced role and social functioning were significantly lower for men with better bowel functioning (role function–physical odds ratio [OR], 0.6, 95% confidence interval [CI], 0.52–0.72; role function–emotional OR, 0.6, 95% CI, 0.49–0.69; social function OR, 0.6, 95% CI, 0.48–0.67). After adjusting for other covariates, a one standard deviation difference in bowel function was significantly associated (P < 0.0001) with a 1.8-point difference in the physical summary scale and a 2.2-point difference in the mental summary scale.
Consistent with the SF-36 results, prostate carcinoma patients affected by reduced bowel function appear to experience a great deal of distress related to this problem. Shown in Table 4, a one standard deviation difference in bowel function is associated with an 18.2-point difference in bowel bother (P < 0.0001). Including other covariates, this model of bowel function explains nearly 60% of the variance in bowel bother.
When we examined the CARES-SF cancer specific outcomes, they too were strongly affected. As shown in Figure 1, the psychosocial scale was most strongly related to bowel function (β = −3.6, P < 0.0001 for a 1 standard deviation difference in bowel symptoms), more so than for sexual or urinary function. The physical aspects of cancer rehabilitation were also more strongly associated with bowel symptoms than sexual or urinary symptoms (β = −2.9, P < 0.0001 for a 1 standard deviation difference in bowel symptoms). A one standard deviation increase in bowel function postcancer diagnosis was associated with a 2.2 point decrease in the global CARES-SF summary measure (P < 0.0001).
Overall, our findings indicated that sexual, urinary, and bowel symptoms after localized prostate carcinoma diagnosis and treatment were significantly associated with nearly all of the generic and cancer specific quality-of-life outcomes examined, including both physical and psychosocial domains. There have been several studies documenting the effect of prostate carcinoma treatment on symptoms associated with various treatment options. However, few have directly examined and compared the effects of sexual, urinary, and bowel symptoms on generic and cancer specific quality of life. Prior studies that have reported results on the association between bowel symptoms and quality of life have not examined the impact across such a wide range of domains.3, 11, 17
Using a single question (After your prostate carcinoma treatment, do you feel your overall quality of life is: better than before/same as before/worse than before?), Kao and colleagues3 found that incontinence and impotence were significant predictors of overall health status (OR, 0.41 and 0.34, respectively for better or same vs. worse). However, no further detail was gathered on the specific domains of quality of life affected in this retrospective study of prostatectomy patients treated at one of five military hospitals.
Using a modified version of the Functional Assessment of Cancer Therapy Scale, Heathcote and colleagues11 found that problems most affecting the quality of life of Australian prostatectomy patients 1–6 years after surgery included impotence (40%), concern about cancer (12%), and bladder problems (8%). The study was also cross-sectional. Like the study by Kao and colleagues,3 no detail was provided on the specific domains of quality of life affected.
One university hospital case series of patients (T1c–T3 classification prostate carcinomas) treated with either radiation or surgery found that problems with sexual function, incontinence, and bowel function were significantly associated with depression, tension, and fatigue.17 This study did not assess generic domains of quality of life, but the researchers did report significant associations between symptoms and the physical domains of the Functional Living Index: Cancer, a cancer specific quality-of-life measure.
A study by Litwin and colleagues of localized prostate carcinoma patients treated in a managed care setting used the same outcome measures as those used in our study to examine the impact of treatment options on quality of life.1 However, our observed mean quality-of-life scores tended to be higher on the SF-36, CARES-SF, and UCLA Prostate Cancer Index than those reported in the study by Litwin et al. 5–6 years after diagnosis. The men in both studies were of similar ages, but notable differences between the studies included the sample population (managed care patients vs. health professionals) and the time since cancer diagnosis (1961–1991 vs. 1993–1998 in our case). It is unclear whether these differences explain the differences in results, but they should be considered when interpreting our findings. In the era of widespread prostate specific antigen screening and more viable treatment options than ever before, quality of life for localized prostate carcinoma patients may have improved over time. However, because better health-related quality of life was observed for both cases and controls in our study compared with the earlier findings of Litwin and colleagues, it is unlikely that advances in prostate carcinoma treatment alone account for all of the difference. More extensive measures of urinary, bowel, and sexual functioning, such as those recently developed by Wei and colleagues37 also may provide further insight regarding the specific quality-of-life impact of various symptoms associated with localized prostate carcinoma.
As a guide to interpreting the quality-of-life impact experienced by the patients in our study, we compared our observed effect sizes to those associated with smoking in the same group of men using multiple regression models of SF-36 summary scales and the CARES-SF summary scale. For instance, the effect of bowel symptoms on the physical summary scale of the SF-36 was as large as the effect of current smoking. The impact of smoking on the mental summary scale of the SF-36 was also comparable to the effect of sexual and bowel symptoms (β = 3.5 vs. β = 3.5, 2.5, and 3.7 for sexual, urinary, and bowel symptoms, respectively). To give another example, the effect of regular exercise (highest vs. lowest quintile) was almost as strong as bowel symptoms for the physical component scale (β = 3.6 vs. 4.0), similar to urinary symptoms on the mental component scale (β = 2.3 vs. 2.5), and approximately half of the effect of bowel symptoms observed for the cancer specific summary scale (β = −2.0 vs. −4.3).
An important limitation of this study concerns the cross-sectional nature of the data. We were able to stratify our results by certain indicators of sexual and urinary function before prostate carcinoma diagnosis. However, because the pre- and postcancer measures were not the same, we were unable to directly analyze the effects of changes in sexual and urinary symptoms on quality of life. We were also unable to determine with certainty whether the symptoms experienced by patients were the result of cancer itself or the associated treatment. Another limitation of this study is the low number of African-American subjects in the cohort. African-American men have the highest incidence and mortality rates of prostate cancer of any group in the world,9 and it is important that the impact of disease and treatment on quality of life be studied in this group. Because of the low number of ethnic/racial minority men included in this study, a great deal of caution should be used in generalizing these results to such men.
In conclusion, we found that sexual, urinary, and bowel symptoms were each associated with nearly all of the quality-of-life outcomes examined, from general health to symptom specific bother. Bowel symptoms after prostate carcinoma had the most severe impact on quality of life compared with urinary and sexual symptoms. Urinary symptoms had a greater impact on quality of life among men with higher levels of precancer functioning. Similarly, sexual symptoms were more strongly related to quality of life among men with some level of sexual activity before prostate carcinoma.
For all localized prostate carcinoma patients in this study, psychosocial/mental quality-of-life domains were the most strongly affected by sexual, urinary, and bowel symptoms. Therefore, it is important for health care providers to address the mental health issues associated with prostate carcinoma when discussing the potential side effects of treatment with patients. Anxiety and depressive symptoms, as well as overall mental well-being, are domains of health that should be considered before making a treatment decision and in the follow-up care of localized prostate carcinoma patients, in addition to physical health concerns. Health care providers also need to be aware that the general physical and mental impact of sexual, urinary, and bowel symptoms may need to be addressed for many cancer patients after treatment for the disease itself has concluded.
- 23SF-36 health survey. Manual and interpretation guide. Boston, MA: The Health Institute, New England Medical Center, 1993., , , .
- 24Measuring health: a guide to rating scales and questionnaires. New York: Oxford University Press, 1996., .
- 26SF-36 health survey. In: MaruishME, editor. The use of psychological testing for treatment planning and outcomes assessment. 2nd ed. Mahwah, NJ: Lawrence Erlbaum Associates, 1999: 1227–46..
- 32SF-36 physical and mental health summary scales: a user's manual. Boston: The Health Institute, 1994., , .