To investigate the incidence of urinary incontinence and its development over time, to compare the effects of alternative definitions on the incontinence rate and to explore risk factors for incontinence after radical retropubic prostatectomy (RRP) for clinically localized prostate cancer.
PATIENTS AND METHODS
Urinary continence was assessed using a questionnaire administered by a third party in 1144 consecutive patients after undergoing RRP at our department from January 1986 to December 2001. Overall, 985 men (86%) were suitable for evaluation (mean age 64.5 years, mean follow-up 95.5 months). We compared the effects of three definitions on the actuarial rate of continence: (1) no or occasional pad use; (2) 0 or 1 pads used daily, but for occasional dribbling only; (3) 0–1 pads daily. The time to recovery of continence was defined as the date on which the patient met the continence definitions. The impact of incontinence on health-related quality of life (HRQoL) was also evaluated. Univariate and multivariate analyses were used to identify predictors of incontinence, using data gathered prospectively.
At the last follow-up at 24 months after RRP, 83%, 92.3% and 93.4% of men achieved continence according to definitions 1, 2 and 3, respectively. The difference in time to recovering continence was significant for definition 1 compared to the others (P < 0.001). Most men using 1 pad/day complained of occasional dribbling only (89.3%), considered themselves continent (98%) and their HRQoL was not as seriously affected as those requiring ≥ 2 pads/day. Men continent (by definition 3) at 2 years had an actuarial probability of preserving continence of 72.2% at the last follow-up. On multivariate analysis the age at surgery (P = 0.009), anastomotic stricture and follow-up interval (both P < 0.001) were independent prognostic factors. Bilateral neurovascular bundle resection was another independent predictive factor (P = 0.03) in the subset of the last 560 men with available data on surgical technique. The reduction in the incidence of incontinence over time was as high as 86%.
Continence improves progressively until 2 years from RRP but some patients can become incontinent later. The criterion of pad use discriminates well between men with a limited reduction in their QoL (no or one pad used) and those with a markedly affected QoL (≥2 pads/day). It could be clinically valid to consider users of 1 pad/day as continent. Age, bilateral neurovascular bundle resection and anastomotic stricture are significant risk factors for incontinence. There was a marked trend for the incidence of incontinence and anastomotic stricture to decrease with time.
University of California Los Angeles Prostate Cancer Index
Urinary incontinence still represents a clinically important complication after open retropubic radical prostatectomy (RRP) for prostate cancer; its incidence ranges widely, from 1% to 47%. To avoid this complication, several authors evaluated modifications to the standard anatomical procedure described by Walsh , e.g. bladder neck preservation [4,5] or intussusception , puboprostatic ligament sparing , cavernosal nerve sparing , seminal vesicle sparing , and control of the deep dorsal vein . Unfortunately, some factors make it difficult to compare the results of different open techniques, or open vs laparoscopic techniques; differences in patient populations (e.g. single institutional-based vs community-based), study size and design, practice setting, experience of the surgeon, time of evaluation, data collection methods (e.g. patient rather than physician reporting [11,12]). Some of the variance in the reported incidence is clearly related to the wide range of definitions of urinary continence used. The most frequent and practical variable used in assessing urinary continence is the daily continence-pad requirement [13–15]. In published studies some authors considered patients continent if using no more than 1 pad/day , whereas others have used the stricter criterion of no pad use [14,15]. There is no agreement when considering continent men using 1 pad/day, which might be affected by a broad range of frequencies of urinary leakage, from patients with good urinary control using 1 pad/day simply as a precaution, to patients with frequent dribbling . Furthermore, the surgeon often influences this variable by discouraging the use of pads in patients with occasional urinary leakage. Therefore, a substantial difference in continence rates might result depending upon the inclusion or not of such patients in the definition of continence.
Thus we aimed to retrospectively evaluate in our surgical series: the incidence of urinary incontinence after RRP by using different definitions based on pad use, as well as its characteristics and impact on health-related quality of life (HRQoL); the time course of recovery of continence (RC) after RRP; the potential role of several clinical variables as risk factors for incontinence; and the temporal trend of this complication over the observation period.
PATIENTS AND METHODS
From January 1986 to December 2001, 1144 consecutive patients underwent RRP (as originally described by Walsh ) at our department for clinically localized prostate cancer (stages cT1–cT3, NXM0). Since 1986 all patient data were prospectively collected in a computerised database. Excluded from evaluation were 159 patients not available or unsuitable for the assessment of continence for the following reasons: dead from prostate cancer or other unrelated causes (139), patient refusal (three), lost to follow-up (13), mental impairment (two) and pad use before RRP (two); thus 985 patients were evaluated.
Table 1 shows the patients’ characteristics and pathological features. No patient received radiotherapy before RRP and 27 (2.7%) had pelvic radiotherapy afterward. A nerve-sparing procedure was used in 238 patients selected on the basis of age, clinical stage, PSA level and sexual function. Pelvic lymphadenectomy (according to Whitmore) was used in all patients; 11 experienced surgeons undertook the surgery. Urethral catheters were removed at 1–3 weeks after surgery if there was no evidence of leakage on cystography. During the period January to June 2005, patients were contacted again for a follow-up and re-evaluation of continence. Tumour stage was assigned using the 1997 TNM system and graded according to the Gleason scoring system. Patients who complained of important urinary voiding symptoms or had significant stress urinary incontinence at 1 year after RRP were evaluated for anastomotic strictures by cysto-urethroscopy. In all, 325 patients had neoadjuvant treatment before RRP and hormonal adjuvant treatment was given after RRP in 341 men.
To assess urinary continence status before and after RRP, a urologist not involved in the surgical procedures or the treatment after RRP directly interviewed the patients (88% by telephone and 12% face-to-face) using a questionnaire including the following: (i) The urinary domain (six items) of the University of California Los Angeles Prostate Cancer Index (UCLA-PCI) , a validated questionnaire capturing incontinence and its impact on HRQoL. The pad requirement was defined as 0, occasional, 1, 2 or ≥ 3 pads/day, rather than 0, 1–2 and ≥ 3/day. The pad requirement was used to categorize the patients, and the remaining five items were linearly transformed in a 0–100 scale, a higher score indicating better HRQoL. The Pearson correlation between pad use and the other items was calculated; (ii) A question capturing the overall self-assessed continence, according to Lepor and Kaci  (‘Do you consider yourself continent?’: yes or no); (iii) Questions capturing incontinence before RRP, the type of incontinence, modification of continence over time, use of drugs and time to RC, recorded as the interval between the date of catheter removal and date on which the patient met the continence definitions (immediately after catheter removal or by 1, 2, 3, 6, 12 and 24 months or later).
We evaluated three definitions of continence: (1) no or occasional pad requirement; (2) no/occasional pad requirement or use of 1 pad/day for occasional dribbling only; (3) the use of 0–1 pad/day (without considering dribbling frequency).
The actuarial and median times to continence, and percentage of continent patients at 3, 12 and 24 months, were calculated using the Kaplan–Meier method for each continence definition. A paired log-rank test was used to assess time-to-event differences among definitions . Using definition 3, we assessed by univariate and multivariate Cox proportional-hazards regression models the association between time to RC and the following variables: age at surgery, preoperative PSA level, history of TURP, clinical stage, neoadjuvant and adjuvant hormone therapy, follow-up interval, intraoperative blood loss, pathological stage, definitive Gleason score, prostatic apex and seminal vesicle infiltration, positive margins and extracapsular extension, clinically apparent bleeding after RRP, and symptomatic anastomotic stricture. A sub-analysis was used to assess the effect of nerve-sparing surgery, i.e. both neurovascular bundles (NVBs) resected, vs at least one preserved, in the last 560 patients (from 1997 to 2001) for which the surgical procedure was systematically recorded. Multivariate analyses were based on significant variables from the univariate analyses. A backward model selection was used to find the meaningful variables related to the event (RC). A two-tailed P < 0.05 was deemed to indicate statistical significance.
Table 2 shows the results of the survey in a cross tabulation examining the relationship between pad requirement and urinary control, and the effect of using alternative definitions. At the last follow-up, 83%, 92.3% and 93.4% of patients achieved urinary continence according to definitions 1, 2 and 3, respectively. Most users of 1 pad/day (89.3%) claimed occasional dribbling only or total control and, as also shown in Table 2, 98% of these patients considered themselves continent.
Table 2. Cross-tabulation of pad requirement vs urinary control, the effect of the different definitions of continence, the self-assessed continence for each pad-requirement group, and the type of urinary leakage according to pad requirement among all patients (364) claiming some urinary dribbling
Daily pad requirement
Ten patients recovered continence after implantation of an artificial urinary sphincter (AMS 800, American Medical Systems, Minnetonka, MN, USA) but were considered as using three or more pads/day and with frequent dribbling.
†Urge leakage was defined as the complaint of involuntary urine loss accompanied by or immediately preceded by urgency; stress leakage as the complaint of involuntary urine loss occurring by an effort or exertion leading an increase in intra-abdominal pressure (cough, sneeze, laugh). The leakage was defined mixed if both stress and urge definition were met.
There was a significant correlation (r ≥ 0.78, P < 0.001) between pad use and the score of any other item of the urinary domain of the UCLA-PCI (Fig. 1). HRQoL was affected in all pad groups but patients requiring ≥ 2 pads/day scored far worse.
Figure 2 shows the actuarial curves of time to RC by definition; the difference in time to RC was significant for definition 1 compared to the others (P < 0.001), whereas there was no significant difference between definitions 2 and 3 (P = 0.32). In subsequent analyses continence was referred to according to definition 3.
The actuarial continence rates at 3, 6, 12 and 24 months were 68.2%, 78%, 86.9% and 93.2%, respectively, and median (sd) time to continence was 4.5 (6) months. The difference between 12 and 24 months was statistically significant (P < 0.001). After 24 months only two patients (0.2%) regained continence. Figure 3 shows the actuarial probability of preserving continence; at the last follow-up, 27.8% (95% CI 5.8–49.8) of the patients continent at 2 years subsequently became incontinent.
Most patients described their urine leakage as stress incontinence (89.6%; Table 2). Urge incontinence alone was infrequent (2%) but 69 patients (7%) reported a previous use of anticholinergic drugs for incontinence and six (0.6%) were still using these drugs. Diuretic use was more frequent among continent than incontinent patients (13% vs 3%; P = 0.03).
Table 3 shows the P values from the univariate and multivariate analyses. Anastomotic stricture, age, NVB resection and follow-up interval were independent negative prognostic factors. A subgroup of 50 patients (5.1%) developed a vesico-urethral stenosis within a mean (range) of 24 (4–24) months of surgery. Patients developing anastomotic stricture had a 2.4-fold higher relative risk (RR) (95% CI 1.7–3.5; P < 0.001) of remaining incontinent. As shown in Fig. 4, RC was better in younger than in older men; e.g. men aged >65 years had a RR of incontinence of 1.23 (95% CI 1.08–1.41; P = 0.0014) compared with younger men. The RC was significantly worse in patients having both NVBs resected (P = 0.030). The incontinence risk was also influenced directly by the follow-up interval, indicating in which period patients had surgery (Fig. 5). The marked temporal trend in the incidence of both incontinence and anastomotic stricture decreasing over the observation period is shown in Fig. 6. The decreasing trend was unrelated to any particular surgeon; one surgeon (F.P.) operated over the whole period and these patients showed a similar trend.
Table 3. Univariate and multivariate analyses of risk factors for incontinence (according to definition 3) after RRP
This study was retrospective and based on a large group of patients with a long follow-up. While the retrospective design of the study is certainly a limitation, it nevertheless offers several clinical advantages. The most important strengths of this study are the large population evaluated, the questionnaire-based methods, the high response rate, the long-term follow-up and the evaluation of different practical definitions of continence. Several clinically significant variables were studied as potential risk factors for incontinence by univariate and multivariate analyses, using data gathered prospectively. Some authors noted that the method of data collection might cause a significant difference in reported incontinence rates [11,12]. It seems that patients tend to minimize complications when speaking to their surgeons and the surgeon-interviewer, in turn, might unconsciously minimize adverse outcomes. To avert this bias, emphasis was placed on the assessment of urinary continence reported by the patient directly to a third party. In the present study, to obtain more objective information, the operating surgeons were not involved with any component of data acquisition or entry.
There was a statistically significant difference in continence rate between the strict definition 1 and the more inclusive definitions 2 and 3. There was no difference between the last two definitions, as there was a small proportion of patients complaining of frequent dribbling among men using 1 pad/day. Mostly these men considered themselves continent and their urinary domain HRQoL scores were meaningfully better than those of men requiring ≥ 2 pads/day. Therefore, we agree with those supporting the contention that men wearing a single pad often do so for reassurance, rather than for significant urinary incontinence, and they can basically be considered continent . Conversely, requiring ≥ 2 pads/day clearly identifies patients seriously affected in their HRQoL. Kielb et al. found in a prospective study of 90 men that no patient requiring 1 pad/day claimed frequent dribbling at 56 days after removal of the urethral catheter, and concluded that ‘the clinical practice of asking patients how many pads daily they use might be valid, as it corresponds well to the impairment they have’. Lepor et al. also found that at 24 months after surgery, all men using 1 pad/day or reporting occasional dribbling considered themselves continent. The same authors reported that definitions of continence requiring total control or no pad use significantly underestimated the patients’ self-assessed continence .
Using the present definition 3, the actuarial 12-month incontinence rate (13%) was close to the corresponding rate reported by Wei et al. (16%) when considering a similar definition. However, the present rate is significantly higher than that reported by other large series (6–8%), despite their stricter criteria of continence [14,15,20]. At 24 months of follow-up the present incontinence rate significantly decreased to 6.8%, in the range of those reported from other academic treatment centres, despite the variable definitions of incontinence and surgical techniques [14,15,21–23]. When considering the same definition in a more recent series (2000–2002) of 500 men, Lepor et al. reported an incontinence rate at 12 and 24 months of 7.9% and 1.5%, respectively; the corresponding rates in the last 2 years of the present series were 4% and 2.7%, respectively. Some authors also reported the incontinence rate at 3 months, at 18–53%[1,18,21,24]; the corresponding rate in the present study was 31.8%.
Our analysis confirmed the results of other reports that the incontinence rate progressively improves through the first 2 years after RRP [1,2,13,22]. Most patients regained continence within 12 months, but another significant proportion (6.7%) did so at 12–24 months. A similar finding (6.4%) was reported by Lepor and Kaci , based on pad requirements. According to the present and other results [1,2,12,18,21,22], we suggest assessing the continence outcome and to consider an invasive treatment for moderate incontinence after ≥ 2 years from RRP. Of interest, continence achieved after surgery can be lost later with ageing. There was a slight (3.5%) worsening in the continence rate at 2–5 years after surgery, assessed prospectively by Penson et al.. Therefore, the incontinence rate at 2 years might underestimate the corresponding rate after a longer follow-up; this finding is expected as the patients age. Moreover, the progressive increase of patients’ drug consumption due to comorbidity might decrease their continence because of drug side-effects.
The age at surgery statistically significantly compromised the continence status; this result concurs with the findings of other studies [21,24,26–28]. Stanford et al. reported that men aged <60 years were significantly less likely to be incontinent at 2 years than were older men. Talcott et al. reported that the continence rate at 12 months was 91% among patients aged <65 years and 85% among those aged ≥65 years at surgery; the present corresponding actuarial rates were comparable (90.8% and 83%). In other studies the age at surgery had no statistically significant effect on urinary control [22,29,30] but, as noted by Eastham et al., these series either included few elderly patients or observed such low rates of incontinence that identification of a significant risk factor in any subgroup would be unlikely.
The most significant risk factor in the present series was the diagnosis of a symptomatic anastomotic stricture after RRP, in agreement with others [21,31]. The incidence of anastomotic strictures after RRP is 0.5–32%[31–34]. The present overall anastomotic stricture rate was 4.5%, but it was much higher when considering incontinent patients only (33.8%). Park et al. reported a similar rate (4.8%) in a series of 753 patients, and found that 62% of men with anastomotic stricture became totally continent and 54% required no protective pad, vs 92% and 88%, respectively, in the control group. Eastham et al. also found that anastomotic stricture significantly affected continence status after RRP in a multivariate analysis (P = 0.015). One study, conducted in a smaller series, reported that this complication had no impact on continence rate .
In the relevant subset of patients analysed there was a significant association between using at least unilateral nerve-sparing surgery and continence after RRP; others reported similar findings [13,21,29,36]. Eastham et al. reported that resection of even one NVB resulted in a substantial decrease in the continence rate. Gralnek et al. noted differences in urinary function or bother between patients undergoing nerve-sparing RRP or with no nerve-sparing. However, this factor is extensively debated; large series failed to find a significant effect of NVB resection [2,14,35–37]. Catalona et al., reporting an update of their results after RRP in a series of 3477 men, confirmed their previous lack of association between nerve-sparing procedure and RC, but the group with no nerve-sparing in their series was rather small (5%). Moreover, it is likely that differences in patient age, tumour stages, fine details of the technique, method of assessment and definition of continence might explain these differences.
As in other reports [14,21], the present multivariate analysis showed no association between continence after RRP and PSA level before, previous TURP, neoadjuvant and adjuvant hormone therapy, intraoperative blood loss, pathological features and clinically evident haemorrhage after surgery. There was a trend for worse continence status with worse pathological characteristics, but this effect was not statistically significant in the multivariate analysis. A few studies have reported an association between a history of TURP before RRP and incontinence afterward [38,39], but this finding was not subsequently confirmed in larger series.
There was a progressive improvement in continence during the present follow-up; this temporal trend is consistent with other reports [21,40]. There was also a parallel trend in the incidence of anastomotic stricture, as noted by others [14,40]. The improved results over time might be explained by patient-related and surgeon-related factors: being older at surgery; more advanced stages were more frequent in the earlier than in the most recent periods; and increasing surgeon and team experience was reported to decrease technical complications over time [14,41]. Continence was unrelated to any one surgeon; the only surgeon (F.P.) that operated over all the period had a similar improving trend as all the other surgeons. Thus it seems that, at a referral centre, team experience is more relevant than the skill of any particular surgeon.
This was a retrospective study and has some limitations. We used interviewing methods to assess the continence status and impact on HRQoL, rather than self-administered instruments. Nevertheless, there was no difference between self-administration and physician-completed instruments when evaluating the International Consultation on Incontinence Questionnaire . A recall bias was probably present in the oldest group of the present patients. Our population encompasses many long-term survivors, but this makes the present results more useful in counselling patients about long-term continence and changes in HRQoL. A prospective trial could give more consistent results, but the present partially retrospective study has the advantage of a long-term follow-up. Our series also contains patients with quite advanced disease (high PSA levels, poor pathological features) than other modern series, mainly due to the contribution of the first 10 years of the observation period. This study was done at one academic centre with high volumes of treated patients; therefore its general applicability to other practice settings could be further limited, even though it might better reflect a target rate of continence for national results.
In conclusion, the present study confirms that the overwhelming majority of patients eventually recover continence after RRP. Most patients using 1 pad/day complain only of occasional dribbling and their HRQoL is not as seriously affected as patients requiring ≥ 2 pads/day. Definitions of continence including users of one pad could be clinically valid. The probability of becoming continent improves progressively after surgery and the final continence outcome is achieved at ≈ 2 years. Age, NVB resection, anastomotic contracture and the era of the operation are variables independently affecting continence after RRP. Our study showed a consistent trend for the incidence of incontinence and anastomotic stricture to improve over time.