DO STROKE REHABILITATION inpatients whose urinary. Urinary Incontinence and Stroke Outcomes. Jan C. Gross, PhD, RN, CS

Transcription

1 22 Urinary Incontinence and Stroke Outcomes Jan C. Gross, PhD, RN, CS ABSTRACT. Gross JC. Urinary incontinence and stroke outcomes. Arch Phys Med Rehabil 2000;81:22-7. Objective: To examine the relation of urinary incontinence to functional status and discharge destination in stroke rehabilitation patients. Design: Historical cohort. Setting: Freestanding, not-for-profit rehabilitation hospital. Participants: Nonrandom sampling. The charts of all admissions to the stroke rehabilitation unit were screened to identify patients with a medical diagnosis of completed stroke occurring 2 to 4 weeks before admission to rehabilitation, documented evidence of urinary incontinence at least daily the first 2 days after admission, and Functional Independence Measure (FIM) bladder sphincter score of 5 or less. Ninety patients met the inclusion criteria. Main Outcome Measures: FIM, change in FIM scores from admission to discharge, weekly FIM gains, and discharge destination. Results: Continence status at discharge was not associated with discharge destination, but did predict total FIM score at discharge and the change in FIM scores from admission to discharge. Subjects who regained continence had higher weekly FIM gains on all subscales except Communication. A critical area of difference between the groups was in Social Cognition scores. The time interval from stroke greatly influenced the findings; in patients evaluated 14 to 18 days after stroke differences between continent and incontinent subjects were found only in Social Cognition scores, whereas in patients evaluated 27 to 30 days after stroke differences were found in Self-Care and Sphincter Control scores. Conclusion: Further investigation into cognitive characteristics should be undertaken with a more comprehensive tool appropriate for an impaired population. Key Words: Urinary incontinence; Outcome assessment (health care); Cerebrovascular accident; Rehabilitation by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation DO STROKE REHABILITATION inpatients whose urinary incontinence (UI) resolves have improved functional outcomes and home discharge rates? Although UI traditionally has been considered an indicator of stroke severity and a predictor of poor outcome, the relation between UI and stroke outcome is From the College of Nursing, University of Kentucky, Lexington, KY. Submitted for publication March 1, Accepted in revised form May 25, Supported in part by National Research Service Award Predoctoral Fellowship Award 5 F31 NR06961 from the National Institute of Nursing Research with additional support from the Office of the Vice-Chancellor of Graduate Studies and Research, University of Kentucky. Presented in part at the 8th Annual Stroke Rehabilitation Workshop, sponsored by Educational Resources, Inc., October 1996, Cambridge, MA, and at the Association of Rehabilitation Nurses Annual meeting, November 1998, Baltimore, MD. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. Reprint requests to Jan Gross, 106 Cherry Drive, Harrodsburg, KY by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation /00/ $3.00/0 unclear. Findings from earlier studies that addressed the relation between UI and outcomes are difficult to interpret and generalize because of heterogeneous samples, differences in time since onset of stroke within a sample, and the use of tools without evidence of reliability and/or validity. 1 Recent research has focused on outcomes measured months to years after discharge from rehabilitation and has not addressed the relation of UI to the rehabilitation process. The present study examined the relation of UI to functional outcomes for stroke patients in rehabilitation and their destination at discharge from rehabilitation while controlling for common methodologic problems. Although UI consistently has been a negative prognostic indicator of outcome after stroke, many early studies supporting this relation have flawed methods. 1 Jongbloed 1 reviewed 33 studies predicting functional recovery based on patient characteristics within the first 3 months after stroke, critiquing studies conducted between 1950 and 1986 that systematically measured functional status and relating these measures to characteristics of stroke survivors. In the five studies that included UI as a factor for evaluation, incontinence was negatively associated with outcome. This relation held when outcome was functional status at discharge 2,3 and when functional improvement was measured. 4 Only two of the five studies addressed the reliability and validity of the outcome measures and only one of the two controlled for the time from stroke. 1 The study that controlled for time from stroke and used tools whose reliability and validity were supported did not measure outcome until 6 months after stroke. 5 The effect of UI on a person s ability to function immediately after stroke onset and its influence on rehabilitation remain largely unexplored. It has been suggested that UI interferes with the rehabilitation process, resulting in less functional recovery. 6 Barer 6 investigated the prevalence of UI, its associated factors, and speed of recovery in 362 subjects in an intervention trial evaluating the use of beta blockers after stroke. Continence and severity of hemiparesis independently predicted the likelihood of survival and return home; continence was the more powerful predictor. Barer found that functional recovery often was faster in continent patients, but acknowledged the difficulty of determining whether this resulted from better adaptation or quicker resolution of cerebral damage. Barer called for further study of the link between UI and outcome after stroke. Recent research that has also supported UI as a strong prognostic factor has not reflected the effects of UI on the acute phase of rehabilitation or addressed time from stroke. Two studies 7,8 evaluated outcome several months to years after hospitalization. Two other studies 9,10 evaluated outcomes in the rehabilitative period and found UI associated with less recovery, lower rate of recovery, and the patient s eventual discharge setting. Although the tools used to measure functional status in these studies have extensive support for validity and reliability with stroke populations, questions arise concerning the time from stroke onset in the comparisons of outcome Continent and incontinent groups are not compared for time from stroke onset in these studies. In the present study the effect of UI on stroke outcomes was examined, controlling for time from stroke onset and using a tool with well-supported reliability and validity with stroke patients. The outcome measures investigated were functional

2 URINARY INCONTINENCE AND STROKE OUTCOMES, Gross 23 status, improvement in functioning, and discharge destination. Four research questions were asked. (1) Does functional status measured at set times from stroke onset differ for patients who regain continence during rehabilitation compared with those who remain incontinent at discharge from rehabilitation? (2) Do stroke patients who regain continence function better by discharge than those remaining incontinent? (3) Does the discharge destination differ according to continence status? (4) Is UI a significant predictor of functional status and level of functional improvement in functioning by discharge from rehabilitation? METHOD Sample Data were collected by means of a retrospective chart review of all 1994 stroke admissions to a rehabilitation hospital. The study site was a freestanding, not-for-profit rehabilitation hospital with a 40-bed inpatient stroke unit that has been in operation for about 20 years and serves approximately 400 stroke patients a year. The inclusion criteria were: (1) medical diagnosis of stroke 2 to 4 weeks before admission to rehabilitation, (2) documented urinary incontinence at least once a day the first 2 days after admission, (3) bladder on the Functional Independence Measure (FIM) of 5 or less on admission, 11 and (4) completion of the prescribed rehabilitation program. A FIM bladder score of 5 was selected to identify patients with any level of UI on admission. Patients who did not complete their rehabilitation program or who developed complications requiring transfer to an acute care hospital or bedrest for more than 3 days were excluded from the study. A sample of 26 patients per group was needed to detect differences between groups with a power of.80 at an alpha of.05, assuming a large effects size. 12 Because a substantial difference is necessary if the difference between the two groups is to be both clinically and statistically meaningful, a large effects size was chosen. Of the 446 admissions in 1994, 90 cases met the inclusion criteria. Slightly more than half (50.8%) of the 356 cases not enrolled were excluded because the patients were continent on admission. Other major reasons for exclusion included stroke occurring more than 4 weeks before admission (18.5% excluded), admission FIM bladder scores greater than 5 (9.8% excluded), and failure to complete the rehabilitation program (8.7% excluded). The majority of the sample were women (64.4%) and white (86.7%). Their mean age was (SD) years (range, 26 to 94yrs). Subjects had multiple medical diagnoses. Urologic diagnoses were recorded in 32.2% of subjects and included urinary tract infection (n 14), prostatic hypertrophy (n 3), and post transurethral prostatectomy (n 7). The majority of cases were unilateral, hemispheric strokes. About three fourths (73%) had no previously documented stroke. Slightly more than half of the sample experienced a right hemispheric stroke; about 37% suffered left hemispheric damage. Sociodemographic characteristics of the two subgroups are outlined in table 1. A detailed description of the sample has been provided elsewhere. 13 Measures Functional status. Functional status refers to an individual s capability or level of skill in performing tasks basic to daily living. The FIM is the functional status component of the Uniform Data System for Medical Rehabilitation (UDSMR). Developed in 1986 to provide a uniform dataset for evaluating rehabilitation programs, services, and outcomes, the FIM Table 1: Characteristics of the Sample as a Whole and by Continence Status Characteristic Total Sample (n 90) Continence Status at Discharge Continent (n 45) Incontinent (n 45) Sociodemographics Mean age (SD), years 70.2 (12.5) 69.4 (14.4) 71.1 (10.7) Gender Female Male Race African-American Caucasian Marital status Single Married Widowed Previous stroke Yes ( 1or ) No Neurologic data Stroke type Infarction Hemorrhage Not Specified Stroke location Right hemisphere Left hemisphere Brain stem Cerebellum Bilateral Not specified Urologic data Urinalysis on admission Positive Negative Indwelling catheter on admission Present None Concurrent urologic diagnosis Yes No Episodes of incontinence First 24 hours, mean (SD) 3.3 (4.1) 2.7 (1.8) 3.8 (2.1) Rehabilitation Length of stay, mean (SD) 29.2 (10.9) 30.8 (10.8) 27.7 (11.0) Days from stroke onset to admission, mean (SD) 14.0 (6.5) 12.1 (5.6) 15.9 (6.7) consists of six subscales measuring self-care skills, sphincter control, mobility, locomotion, communication, and social cognition. 14 The patient s ability to perform each activity is scored on a 7-point scale with each level of the scale representing gradations of independent and dependent behaviors. Scores for each item within a subscale are added to make a subscale score, and subscale scores are added to produce the total FIM score. Scores range from 126 for total independence to 18 for complete dependence. FIM subscales scores also are divided into motor subscales (self-care, sphincter control, locomotion, and mobility) and cognitive domains (communication and social cognition subscales). The reliability and validity of the FIM for use with stroke patients are well documented An

3 24 URINARY INCONTINENCE AND STROKE OUTCOMES, Gross interrater agreement of 80% is required annually by the distributor of the FIM for all facilities using the scale. Improvement in function. Improvement refers to changes in functional status from admission to and discharge from rehabilitation and provides a measure of progress. The UDSMR uses the FIM weekly gain as a measure of the effectiveness and efficiency of rehabilitation. 18 The gain score is obtained by averaging the change in a patient s FIM score from admission to discharge over the length of the patient s stay in rehabilitation. The change in FIM scores from admission to discharge and FIM weekly gain score were the measures used in this study to assess progress in functioning during rehabilitation. Discharge destination. Discharge destination refers to the setting the subject was discharged to when she or he left the rehabilitation facility. The categories for discharge destination were: own home, home of other, and long-term care. Continence status. Continence status was based on evaluations of patients at staffing conferences, in nursing notes, and daily care records (including intake and output charts). The frequency of incontinent voids and the total number of voids were collected the first 24 hours (day 1), the second 24 hours (day 2), and the 24 hours preceding the day of discharge. Continence status at admission for patients with indwelling urinary catheters or in urinary retention requiring intermittent catheterization was determined after the catheter was removed or voluntary voiding began. In those cases where incontinence continued on an intermittent basis (less than daily) throughout the stay, continence status for the 48 hours before discharge was the deciding factor for continence status at discharge. (Only four subjects exhibited intermittent incontinence throughout their stay, and because none were incontinent during the last 48 hours in rehabilitation, all four were categorized as continent at discharge.) Data on demographic and neurologic characteristics also were collected to compare those who regained continence by discharge and those who did not. Factors that may influence the outcome and analysis were: age, gender, marital status, location of stroke, history of previous stroke, length of stay, and concurrent medical conditions. Procedure The study was approved by the university s Medical Center Institutional Review Board and the Research Committee of the rehabilitation facility. A computer-generated master list of all stroke admissions from January 1, 1994 to December 31, 1994 was obtained. Admission records, FIM summary sheets, and staffing reports were the major sources of data. The stroke program manager evaluated records for inclusion in the study and completed the data collection tool on 10 randomly selected charts to check the reliability of chart abstraction. Investigators agreed on the inclusion of cases checked and the data records. Data Analysis Characteristics of the total sample at admission were examined descriptively. The sample was then divided into two groups, one consisting of subjects who had regained continence by discharge and the other containing those who were incontinent at discharge from the rehabilitation program. The two groups were compared using Wilcoxon rank sums for differences in age, time from stroke to admission and to discharge from rehabilitation, length of stay in rehabilitation, and level of UI on admission. Total FIM and subscale scores upon admission and at discharge were compared using Wilcoxon rank sums to determine if there were differences between the two groups. Functional status scores of incontinent and continent subjects, as measured by the FIM, were also compared with Wilcoxon rank sums at two set time intervals from stroke onset. Chisquare was used to evaluate the association of UI with discharge destination. Progress in rehabilitation was determined by calculating the change in FIM scores from admission to discharge within each group, then compared across groups using t tests. The weekly FIM gains were calculated to determine changes in trends or patterns of scores during the rehabilitation program. Using multiple regression, the influence of continence status on discharge total FIM scores and change in FIM scores was studied. Continence scores within the FIM instrument were removed from the total FIM and change score calculations for this analysis. RESULTS Forty-five eligible subjects for each group were identified in the chart review, although the equal number was not intended. The two groups did not differ on sociodemographic variables, neurologic characteristics, or concurrent medical conditions. 13 Functional Status The groups differed on time from stroke onset to rehabilitation admission. Stroke patients who regained continence were admitted to rehabilitation earlier than those who remained incontinent. 13 Time intervals from stroke were selected to compare functional status scores for the continent and incontinent groups. The time intervals were 14 to 18 days and 27 to 30 days after stroke, intervals that corresponded to the sample s average time from stroke to admission and from stroke to discharge, respectively. Of the patients, 43 were assessed 14 to 18 days after stroke (18 incontinent, 25 continent) and 35 were evaluated 27 to 30 days after stroke (23 incontinent, 12 continent). Comparisons revealed that 14 to 18 days after stroke, the only differences in functional scores between the continent and incontinent subjects were in Social Cognition ( p.01). At 27 to 30 days poststroke the Self-Care ( p.05) and the Sphincter Control ( p.001) scores differed by continence group. For both comparisons the incontinent patients scored lower on the identified subscales (figs 1 and 2). These results contrast with what was found when we compared the two groups using admission and discharge scores as the reference point for the comparisons. The group that remained incontinent scored lower on the admission FIM subscales, with the exception of Locomotion and Communication. The total admission FIM score was lower for persons incontinent at discharge than for those who were continent, and these differences remained at discharge. Although the incontinent group improved on all measures, they still had lower mean FIM scores at discharge on all subscales except Communica- Fig 1. Comparison of FIM subscale scores 14 to 18 days poststroke (n 43):, incontinent;, continent.

4 URINARY INCONTINENCE AND STROKE OUTCOMES, Gross 25 Table 2: Multiple Regression Analyses of Predictors of FIM Scores at Discharge for Stroke Patients (n 90) Predictor Standardized Beta t(86) Indices Index F(1, 86) FIM admission score * * Interval from stroke to admission Continence status at discharge * * Sphincter subscale scores deleted from total FIM scores for this regression. * p.001. Fig 2. Comparison of FIM subscale scores 27 to 30 days poststroke (n 35):, incontinent;, continent. tion, 13 which did not differ between groups in any of the comparisons. Rehabilitation Progress To compare progress in rehabilitation, FIM score changes between admission and discharge were examined. Both groups improved in functional status scores during rehabilitation. Those who regained continence had greater increases in five of six FIM scores than those who were incontinent at discharge. Figure 3 shows the differences in admission and discharge FIM scores for the two groups. The mean weekly FIM gains for the continent and incontinent group were compared using t tests. The continent group had higher weekly FIM gains on all subscales except Communication. These differences are reflected in the results comparing motor domain scores and cognitive domain scores. Those who regained continence by discharge had higher weekly FIM gains for motor scores (t(69) 3.8; p.0003). Cognitive gains weekly did not differ by continence status at discharge, (t(66) 1.7; p.10). The inability of the FIM to detect cognitive differences may be attributable to absence of differences in communication or may reflect limitations in the FIM s sensitivity to cognition. Continence Status as a Unique Predictor Using ordinary least squares multiple regression, FIM discharge scores were regressed on FIM admission scores, interval from stroke to admission, and continence status. Sphincter scores were deleted from the FIM values for this analysis. The equation containing these three variables accounted for 71% of the variance in the total FIM score at discharge (F(3, 86) 73.45, p.0001). Beta weights and uniqueness indices (table 2) were reviewed to assess the relative importance of the three factors in predicting FIM scores at discharge. The uniqueness index for a given prediction is the percentage of variance in the criterion accounted for by that predictor, beyond the variance accounted for by other variables. Only FIM admission scores and continence displayed significant beta weights. The FIM admission score displayed the strongest association with the FIM score at discharge. Incontinence at discharge was associated with lower FIM scores at discharge. The finding regarding uniqueness indices matched those for beta weights, in that only the admission FIM score and continence status displayed significant indices. The admission FIM score accounted for 44% of the variance in discharge FIM scores, whereas continence accounted for only 6% of the unique variance. Similar analyses were performed by regressing changes in FIM scores over time onto admission FIM scores, time from stroke to admission, and continence status at discharge. These three variables accounted for 17% of the variance in FIM score change (F(3, 86) 7.21, p.0002). The beta weights and uniqueness indices are listed in table 3. Only continence status was a significant predictor of change in FIM scores. Continence status alone accounted for 17% of the variance in FIM score change (F(1, 86) 20.77, p.001). Discharge Destination Discharge destination did not differ by continence status ( ). The majority of subjects in both groups returned home at discharge from the rehabilitation program. Sixty-four percent (n 29) of continent and 42% (n 19) of those Table 3: Multiple Regression Analyses of Predictors of Change in FIM Scores at Discharge for Stroke Patients (n 90) Fig 3. Comparison of admission and discharge total FIM scores for stroke patients in rehabilitation by continence status over time (n 90):, continent;, incontinent. Predictor Standardized Beta t(86) Indices Index F(1, 86) FIM admission score Interval from stroke to admission Continence status at discharge * * Sphincter subscale scores deleted from total FIM scores for this regression. * p.001.

5 26 URINARY INCONTINENCE AND STROKE OUTCOMES, Gross incontinent at discharge were discharged to their own homes. Another 13% (n 6) of the continent group and 24% (n 11) of the incontinent were released to the home of family or friends. Approximately one fourth (22%) of the continent and one third (33%) of the incontinent were transferred to long-term care. DISCUSSION This study s findings support the association of continence status and poorer poststroke outcome identified in previous research. 1,10,19 Subjects whose UI persisted throughout their rehabilitation program exhibited lower functional status at discharge and made less improvement in rehabilitation. Persistent incontinence was not the only relevant factor associated with poor functional outcomes, however. For this sample, admission FIM scores had a strong, positive relationship with discharge scores as discovered in other stroke outcome research. The role and relative importance of recovery versus adaptation as a principal factor in functional scores remains unclear, as does an answer to the question whether the presence of UI interferes in rehabilitation process or indicates more severe stroke. The common belief that the presence of UI is a primary reason for placement in long-term care did not hold true in this sample. UI did not influence discharge destination. Interpretation of this finding, however, must include consideration for the family support required for rehabilitation admission and the realization that the findings reflect discharge destination from the facility. The possibility of placement into long-term care from home as a result of the burden of caring for an incontinent person over time would not be reflected in these results, nor would the return home of persons initially discharged to long-term care. The results of this study suggest that incontinent patients are at greatest risk of making the least improvement in rehabilitation. The development of more aggressive and progressive treatment of UI early in rehabilitation may alter the adverse outcomes associated with incontinence. Studies evaluating the effects of bladder management in stroke patients have been minimal and have focused on urologic parameters, not on the impact of UI on overall functional status The influence on outcome of treating incontinence, or how continence might be achieved, has been minimally studied. 23 The present study has limitations in two main areas: retrospective data collection and lack of detail in the FIM. The lack of UI history and symptoms on patients charts prohibited examination of specific types of incontinence and limited our exploration of UI type and outcome. Our dependence on FIM scores as the only measure for the effects of stroke prevented analysis of the effect of stroke severity and specific deficits on the presence and persistence of urinary incontinence and on outcomes. The FIM focuses on the performance of functional activities, not on specific deficits. Information on specific deficits was not included in the data collected. Other features of the FIM limited the analysis and interpretation of data. The FIM subscales contain a minimum number of items for each domain of rehabilitation measured. Although the FIM includes questions on communication and cognitive functioning, it has only two items for each of these domains. The sensitivity to and specificity of the FIM in those domains is limited by the small number of items used to capture these complex areas of functioning. These limitations may have biased the results of the study toward findings of no differences between the groups in these areas. Further investigation into cognitive characteristics should be undertaken with a more comprehensive tool appropriate for an impaired population. Future research should be directed toward developing behavioral interventions for use with persons who have had a stroke and evaluating the effectiveness of the approaches on UI and outcomes. To guide the development of interventions, comprehensive information on cognition, the nature of UI and urologic functioning, and their relationships to stroke severity and outcomes are needed. Clinical trials comparing treatments will further our understanding of the relations between UI and stroke outcome and the theoretical mechanisms underlying the interventions. Acknowledgments: The author thanks Margaret Grier, RN, PhD, and Lynne Hall RN, DrPH, for their critical review of the study proposal and manuscript and Mary Kay Rayens, PhD, Biostatistics Consulting Unit, University of Kentucky for her critique of the project and data analysis. References 1. Jongbloed L. Prediction of function after stroke: a critical review. Stroke 1986;17: Wade D, Skilbeck C, Hewer R. Predicting Barthel ADL score at 6 months after an acute stroke. Arch Phys Med Rehabil 1995;64: Wade D, Wood B, Hewer R. Stroke: influence of patient s sex and side of weakness on outcome. Arch Phys Med Rehabil 1985;65: Jimenez J, Morgan PP. Predicting improvement in stroke patients referred from inpatient rehabilitation. CMA J 1979;121: Wade DT, Hewer R. Outlook after stroke: urinary incontinence and loss of consciousness compared in 532 patients. Q J Med 1985;221: Barer DH. Continence after stroke: useful predictor. Age Ageing 1989;18: Taub NA, Walfe CDA, Richardson E, Buraey PGI. Predicting the disability of first time stroke sufferers at 1 year. Stroke 1994;25: Nakayama H, Jorgensen HS, Pedersen MA, Raaschou HO, Olsen TS. Prevalence and risk factors of incontinence after stroke. The Copenhagen Stroke Study. Stroke 1997;28: Owen DC, Getz PA, Bulla S. A comparison of characteristics of patients with completed stroke. Those who achieve continence and those who do not. Rehabil Nurs 1995;20: Ween JE, Alexander MP, D Esposito M, Roberst M. Incontinence after stroke in a rehabilitation setting: outcome associations and predictive factors. Neurology 1996;47: Uniform Data System for Medical Rehabilitation. Buffalo (NY): Research Foundation, State University of New York at Buffalo; Cohen J. A power primer. Psychol Bull 1992;112: Gross J. A comparison of the characteristics of incontinent and continent stroke patients in a rehabilitation program. Rehabil Nurs 1998;23: Granger CV, Hamilton BB, Keith RA, Zielenzy M, Sherwin FS. Advances in functional assessment for medical rehabilitation. Top Geriatr Rehabil 1986;1: Dodds TA, Martin BP, Stolov WC, Deyo Ra. A validation of the functional independence measurement and its performance among rehabilitation inpatients. Arch Phys Med Rehabil 1993;74: Hamilton BB, Laughlin JA, Fiedler RC, Granger CV. Interrater reliability of the 7-level functional independence measure (FIM). Scand J Rehabil Med 1994;26: Gresham GE, Duncan PW, Stason WB, Adams HP, Adelman AM, Alexander DN. Post-stroke rehabilitation. Clinical practice guideline. AHCPR publication no Rockville (MD): Agency for Health Care Policy and Research, Public Health Service, US Department of Health and Human Services; 1995.

6 URINARY INCONTINENCE AND STROKE OUTCOMES, Gross Granger CV, Ollenbacher KJ, Fiedler RC. The uniform data system for medical rehabilitation. Report of first admissions for Am J Phys Med Rehabil 1995;74: Gelber DA, Good DC, Laven LJ, Verhulst SJ. Causes of urinary incontinence after acute hemispheric stroke. Stroke 1993;24: Gross J. Bladder dysfunction after stroke: it s not always inevitable. J Gerontol Nurs 1990;16: Marks RL, Bahr GA. How to manage a neurogenic bladder after stroke. Geriatrics 1977;32: Middaugh SJ, Whitehead WE, Burgio KL, Engel BL. Biofeedback in treatment of urinary incontinence in stroke patients. Biofeedback Self Regul 1989;14: Brittain KR, Peet SM, Castleden CM. Stroke and incontinence. Stroke 1998;29:524-8.