Objectives: To provide a descriptive overview ofthe elderly, nursing home patient population withurinary incontinence (UI).
Methods: This study was a descriptive, cross-sectionaldatabase analysis (2002-2003) examining UIprevalence, demographic and clinical characteristicsof UI patients, and UI pharmacotherapy prevalencein the nursing home setting.
Results: Of the 29 645 eligible subjects, 8995experienced some level of UI at the time the minimumdata set (MDS) was completed (30%).Compared with continent residents, a greater percentageof incontinent residents were older, whitewomen and had a longer length of stay. Incontinentresidents also had more indicators of frailty thanthose who were continent; they were more impairedon activities of daily living and cognitive performancescale scores, were hospitalized more frequently, andhad more urinary tract infections, pressure ulcers, anddepression. More incontinent residents were usingpads/briefs and had bladder retraining and scheduledtoileting. Only 8.7% of those residents rated as havingthe most severe level of incontinence (MDS level 4)were being treated with pharmacotherapy. Of the8995 residents with a UI rating of 1 to 4, only 8% (n =731) had pharmacotherapy.
Conclusion: There is a high prevalence of UIamong nursing home residents and having thiscondition is negatively correlated with measuresof resident health status and healthcare utilization.A variety of interventions are used in this settingto treat UI, and use of pharmacologic therapyappears to be quite low. Appropriate use of interventionalstrategies that may include drug treatmentfor UI in the nursing home may reduce thesubstantial personal and cost burdens associatedwith this condition. However, clinicians may needpopulation-specific scientific evidence in determiningwhich nursing home patients will benefit most frompharmacotherapy.
(Am J Manag Care. 2005;11:S112-S120)
Urinary incontinence (UI) is a significantproblem in elderly populations.The prevalence and economic burdenof incontinence are difficult to estimatesince many sufferers never seek treatment.1,2 It has been reported that 17% to 55%of community-dwelling elderly, and as manyas 50% to 70% of elderly nursing home residentsmay suffer from this condition.3,4Incontinence has been identified as a riskfactor for nursing home placement after hospitaldischarge.5 The high prevalence of UI inthe nursing home population may not indicatethe cause of the facility admission perse, but is likely to be related to the manyconditions that are associated with UI forwhich the elderly require care (eg, functionaldisability, infections, dementia, andneurologic disease). The sequelae of UI, particularlywhen poorly managed, include skinirritation, pressure ulcers, falls, and fractures.UI also has a devastating impact onself-image and quality of life.4
In a review, Hu et al reported estimates ofthe annual direct costs of UI in the UnitedStates as $19.5 billion, with $5.3 billionassociated with care for institutional residents.6 Wilson et al similarly reported annualdirect nursing home costs to be $5.5billion.3 One study included both direct andindirect costs for UI in all persons 65 yearsof age and older in the United States; itreported total costs at more than $26 billionin 1995, with institutional care (includinghospital care) at more than $8.4 billion.7Others predict that in the year 2025 theestimated 8.5 million nursing home residentswill generate $25 billion in direct UIcosts.8 Similarly, estimated costs for overactivebladder (OAB), which is related to andincludes many UI patients, has been estimatedin the United States to be about $3.5 billionannually for institutional residents.6
The care of an incontinent nursing homepatient is significantly more expensive thanfor a continent one. Shih et al found that theincremental labor costs associated with caringfor an incontinent patient are an additional$4957 per patient per year more thanthe costs for caring for a continent patient.9This estimate included the costs of nursingand aide time for changing diapers andlinens, turning and positioning patients,checking for wetness, and assisting patientsin getting to the bathroom. However, itshould be noted that this study also includedthe costs of patients with UI who mayhave also had bowel incontinence.
Treatment of UI typically begins withnoninvasive, conservative approaches.4Nonpharmacologic interventions includebehavioral therapy, including patient education,fluid management, bladder retraining,pelvic floor exercises, biofeedback,and timed bladder emptying. However,patients should be cognitively intact. Inaddition, surgical procedures and catheterizationare appropriate for some patients.There are also emerging technologies thatinvolve electrical stimulation of pelvicmuscles.10
Pharmacotherapy is effective in manypatients and usually is recommended inaddition to behavioral modification whenthat measure alone fails.11 The antimuscarinicmedications, tolterodine and oxybutynin,are the most commonly prescribeddrugs.12 Pharmacologic therapy is a primarymode of treatment for community-dwellingUI patients, and the trial of such therapiesunder appropriate circumstances has beensuggested as a quality indicator in the nursinghome setting.13 However, drug therapy isreported to be underused in the nursinghome.4 It is speculated that the reasons forthis include a lack of efficacy data in thenursing home population, no studies regardinguse of drugs in patients with severe cognitiveimpairment who have likely failedbehavioral therapy, and the risk of contraindications.4
There is a high prevalence of and costburden for UI in the nursing home setting.Underuse of pharmacologic interventionsis reported despite patient preferences andrecommendations for treatment.4,13,14 Yet,little is known about UI, its comorbiditiesand sequelae, and the use of drug treatmentin this care setting in the UnitedStates.15,16 Although the proliferation ofnew treatments has resulted in increasedinterest in UI and OAB overall, thereappear to be no descriptive reports in thepublished literature on prevalence of UIpharmacologic treatment for nursing homepatients with these therapies. The objectiveof this study was to provide a descriptiveoverview of the elderly, nursing home UIpatient population, including demographicsand associated conditions, with a specificfocus on prevalence of drug therapy for thetreatment of UI.
This study was a descriptive, cross-sectionaldatabase analysis examining prevalenceof UI, demographic and clinicalcharacteristics of UI patients, and prevalenceof drug use for UI by residents in thenursing home setting.
Multicenter data from 378skilled nursing facilities (SNFs) from acrossthe United States were used for this analysis.The data include comprehensive informationon more than 200 000 residents nationwidewith varying demographics. It has datafrom a variety of sources including minimumdata set (MDS) assessments, nursingprogress notes, related care plans, medicationand treatment records, physicians'orders, accounts payable/receivable, andcost accounting.
All nursing homepatients were eligible if they were admittedand received care in 1 of the 378 participatingnursing homes in the United Statesbetween January 1, 2002, and December 31,2003. Residents were not eligible for inclusionif they were receiving hospice care,were comatose, or had any level of bowelincontinence during the period as indicatedon the MDS (Section H, question 1.a). Also,any patient with a short stay of 14 days orless was excluded.
The MDS question onbladder incontinence (Section H, 1.b) wasused to classify patients on level of UI, with0 (no incontinence) to 4 (incontinent mostof the time). The maximum level found onany available MDS completed during thestudy period was used. That highest-levelMDS was the source for the data used in thestudy. Each resident was assured of having aminimum of 1 MDS completed as all wereadmitted during the study period and anMDS must be completed on admission byregulation. Additional MDS assessments mayhave been completed after admission, as atleast 1 is required annually (more often insome states) and 1 is required on a significantchange in health status.
Residents were considered to have hadpharmacologic treatment for UI if they wereprescribed any of the following medicationsduring the follow-up period: tolterodineimmediate or extended release; oxybutyninimmediate, transdermal, or extended release;desmopressin; or flavoxate. Patients who didnot receive any of these medications wereconsidered not pharmacologically treated.
Because this was a cross-sectional analysis,it was not possible to discern the timesequence of comorbid events and outcomesto distinguish existing conditions, risk factors,or the sequelae of UI. Thus, these conditionsare referred to here as associatedconditions. Study data were grouped into 3categories: (1) Patient demographics: age,sex, ethnicity, and length of stay (LOS); (2)Associated conditions and other indicatorsof patient health status: activities of dailyliving (ADLs), average time involved incare, cognitive performance score (CPS),all-cause hospitalization, urinary tractinfection, insomnia, pressure ulcer, depression,use of wheelchair for locomotion, andbedfast; and (3) Use of incontinence-relatedprogram or device: bladder retraining,pads/briefs, and scheduled toileting. Forexample, the following measures werederived from the MDS assessments in thefollowing ways:
This was an exploratorystudy that was not intended to testany specific hypotheses, but was intendedto gather hypothesis-generating informationon this population. As such, there was noadjustment made for examining multiple endpoints. Bivariate analyses, stratified by UIlevels and treatment group, were conductedto evaluate resident demographics, associatedconditions, and clinical characteristics.Primary analyses were unadjusted, and descriptivestatistics (proportions, means, and95% confidence intervals of the means, asappropriate) were prepared to compare differentpatient groups. Associated conditionswere evaluated, stratified by themaximum bladder-continence levelachieved by the patient during his/her stayin the SNF and by treatment with drug (orno drug treatment).
Analyses of covariance (ANCOVA) methodswere employed to adjust for covariatesthat were found statistically significantly differentin bivariate analyses. The ANCOVAanalyses were conducted with ADL score,average time involved in activities, CPS, presenceof pressure ulcers, and all-cause hospitalizationsas the dependent variable. Otherstudy end points (significant in the bivariateanalyses) were used as control variables (eg,demographics, associated conditions, andother indicators of patient health status). Tworegression models were examined, one controllingfor pharmacological treatment (treatedvs untreated) and the second for UI level.
A total of 87 000 residents had at least 1MDS assessment available during the studyperiod, 2002 and 2003. Of those residents,29 645 met the inclusion criteria. It shouldbe noted that 27 951 patients were excludedbecause of having had some level of bowelincontinence, alone or in conjunction withUI, which was indicated on their MDS duringthe study period. Before excluding thosepatients with bowel incontinence, the prevalenceof UI was 58% (33 415 of 57 596).
However, among the 29 645 eligible forthis study, 8995 were experiencing somelevel of UI at the time their MDS was completed,a prevalence of 30%.
Results are presented in Tables 1 through4 for the 29 645 patients who met the inclusioncriteria. As expected, most patientswere elderly, had a mean age of 78 years(standard deviation 13.2 years, range 20-109), and were women (63.8%) (Table 1).Patients were mostly white (89.2%) and hadan LOS of 116.3 days during the study period.Those with more severe UI tended to beolder, female, and had the longest LOS.
Across the UI level groups, the rate ofdrug treatment is presented in Table 2.There were 1461 patients treated with UIpharmacotherapy. Of those pharmacologicallytreated, 730 patients (50%) had notexperienced any incontinence during thestudy period (ie, the highest level of UIindicated on any MDS for that period waszero). Age was not significantly differentamong the treated versus untreatedgroups. More of those treated werewomen (76.7% treated vs 63.2% untreated,<.002) and had significantly longerLOS (134.2 days treated vs 115.3 untreated,<.001).
Other conditions or indicators of healthassociated with UI are listed in Table 3. Moresevere levels of UI were associated withmore severe ADL impairment, more caretime needed, more severe cognitive impairment,more frequent hospitalizations, moreinsomnia, more frequent pressure ulcers,greater use of a wheelchair for locomotion,and more bedfast patients (².001 for alltrends). Those pharmacologically treatedwere significantly more impaired on ADLsand cognitive performance (<.001 and= .004, respectively) and had more depression(<.001) than those untreated.However, those untreated experienced morehospitalizations and pressure ulcers (<.001and .02, respectively).
For UI-related programs and devices(Table 4), the more severe levels of UI wereassociated with more use of bladderretraining, pads/briefs, and scheduled toileting(<.001 for all). Treated groups madegreater use of these programs and devices.Only pads/briefs and scheduled toiletingwere significant.
An ANCOVA involving 28 504 patientsexamined various measures adjusted fordemographics, comorbid conditions, modesof locomotion, and incontinence appliances/programs. Of these patients, 1347were treated pharmacologically and 27 157were not treated with a drug. The measuresand overall means for all patients were: ADLscore (14.76), rating of average timeinvolved in activities (1.02), CPS score(1.21), percentage of hospitalized patients(18.9%), and percentage of patients withpressure ulcers (22%). No significant differencesbetween treatment groups were foundwith respect to the adjusted mean score forADL (14.69 vs 14.84, respectively; = .43),rating of average time involved in activities(1.01 vs 1.02; = .14), mean cognitive performancescore (1.19 vs 1.23; = .23), andpercentage of patients with pressure ulcers(20.2% vs 23.9%; = .07). However, thosetreated pharmacologically had a lower prevalenceof hospitalization compared withthose not treated with a drug (15.1% vs22.7%; <.001). Results of the secondANCOVA (adjusted for demographics, comorbidconditions, modes of locomotion, and UIappliances/programs) indicated that UI level(maximum level of incontinence during thestudy period) remained significantly associatedwith ADL score, CPS score, all-causehospitalizations, and prevalence of decubitusulcers (<.001).
In a cross-sectional analysis that examinedresidents at one point in time, it is notpossible to assess causality; it is only possibleto identify associations. For example, inthis cross-sectional analysis, it is not possibleto assess the effectiveness of drug treatmentat various levels of UI because changeover time was not measured (among other unmeasuredvariables). Cross-sectional analysesare useful because they can be a low-cost,easy-to-execute snapshot that informsfuture research and hypothesis generation.With a paucity of data to inform research onUI drug treatment in the nursing home population,the data presented here can be usedto refine future analyses and generate specificresearch questions on this populationin this setting.
As expected, UI residents when comparedwith continent residents were found to beolder, white women with a longer LOS(Table 1). Those residents with UI werefound to have indicators of more frailty thanthose without. UI residents of nursinghomes were more impaired on ADLs andCPS, were hospitalized more frequently,and had more urinary tract infection, pressureulcers, and depression than continentresidents (Table 3). As expected, more UIresidents were using pads/briefs and hadbladder retraining and scheduled toileting(Table 4).
Treatment for UI with pharmacotherapyappears to be quite low. Only 8.7% of thoseresidents rated as having the most severelevel of UI on their MDS (level 4) were beingtreated with pharmacotherapy. Of the 8995residents with a UI rating of 1 to 4, only 8%(n = 731) had pharmacotherapy. The bivariateanalysis indicated that those being treatedhad more impairment on ADLs and CPS,more urinary tract infection, and moredepression, yet, they had fewer hospitalizationsand pressure ulcers. ANCOVA resultsindicated that differences between treatmentgroups on ADLs, CPS, and pressureulcers were not significant when adjusted bydemographic and health characteristics.However, differences in hospitalization persistedafter adjustment.
This may indicate that when deciding toselect pharmacotherapy for a resident, cliniciansare making conservative choices.They may be reserving therapy for themore severe cases. However, this may alsoindicate that there is a threshold of residentfrailty that is factored into the clinician'sdecision. Once a resident crossesthat threshold, pharmacotherapy may beviewed as too risky. For example, pharmacotherapymay be selected for a level 3 or 4resident with impairment of ADLs, but whootherwise has few long-term medical conditionsthat put him/her at risk for hospitalization.However, pharmacotherapy wouldnot be selected for a level 3 or 4 residentwith multiple, chronic comorbid conditions.This is a hypothesis that should betested with further research.
Ouslander et al conducted a trial thatcompared a UI management program withand without tolterodine.18 They found thatthe addition of the drug increased drynessrates by 29% in residents who were clinicallystable. However, only 22% of residents metthe trial criteria of being clinically stable inthat study.4,18 Rates may have been differentif a less restrictive population had been eligible.Thus, the results found here, low druguse and use that appears to be for severe butlow-risk UI residents, may reinforce thatthere is physician uncertainty regarding theappropriateness of pharmacotherapy acrossa range of patients with varied illnesses andfunctional abilities. Before clinicians can feelmore comfortable with these decisions, evidenceof safety and efficacy in this populationwill need to be incorporated into anaccepted clinical approach (eg, clinical protocolsor guidelines).
In general, gaps in care and opportunitiesfor improvement exist in UI diagnosisand care delivery in nursing homes.19,20Clearly, an assessment is needed to identify reversible causes and to make optimaltreatment choices. Studies indicate that supportivebehavioral programs work, but arenot widely used as recommended or asmeasured in nursing home indicators.20,21Based on emerging clinical evidence thatdrug therapy can improve results more thanbehavioral therapy alone,18,22 more specificpharmacotherapy recommendations can beincorporated into or be elaborated on inexisting protocols and indicators.13 However,caregivers and clinicians must be convincedthat the processes and drugs work and mustbe motivated to utilize them. In addition,patients must be cognitively intact for thesetreatments to be effective.
The barriers to pharmacotherapy use forUI in the nursing home include lack of evidenceof safety and efficacy in the population,clinical uncertainty about the benefit-riskbalance in frail elderly, and an attitude thatUI in elderly people is expected.4,23 In a qualitativestudy, Robinson23 found that nursinghome residents have come to believe that UIis inevitable, and in that context, theyattempt to protect their physical, psychological,and social integrity. Robinson states,"Given such beliefs, it is not surprising thatresidents invest more in protecting themselvesfrom the consequences of leakage thanin seeking treatment for UI." In addition,most nursing interventions (eg, frequent toileting,behavioral programs) may include anassessment of drug therapy to evaluate drugsas the cause for UI, but the majority do notinclude recommendations for suggesting drugtreatments.
Johnson et al conducted a study that comparedtreatment preferences of older adultresidents of care facilities (cognitively intactnursing home and residential care residents),nursing staff caring for UI patients, and familymembers of UI nursing home residents.14In paired choices that included diapers,catheters, prompt voiding, and electricalstimulation, the respondents most commonlyselected the least invasive choices.Interestingly however, older respondentsshow a significant preference for medicationsover diapers (<.0001) when compared withnurses and family members who preferreddiapers to medications. Seventy-seven percentof older respondents preferred medicationsto diapers; however, only 21% of familymembers and 39% of nurses preferred medicationsto diapers.14
Thus, the low use of pharmacotherapy forUI in the nursing home is multifaceted. Toensure the best treatments for elderly residents,research on this population ofpatients must be translated into educationalinformation for prescribing physicians whowant to be assured of safety and efficacy, fornursing staff to identify best practices andprocesses, and for patients who need understandabledescriptions of the condition andtreatment options.
Although no specific hypotheses weretested here, it was expected that this populationwould be similar to those evaluated inother nursing home reports. Using MDS datafrom 1990, Brandeis et al examined basiccharacteristics of nursing home residentsbefore the availability of the UI drugs usedtoday; thus, pharmacotherapy was not studied.16 However, it is interesting to note similaritiesand differences in characteristics. Inthe Brandeis et al study, residents were olderthan those identified here (>83 years vs 78years), more were women (>75% of the samplevs 64%), and race was comparable (86% ofthe sample vs 89%).16 Differences may becaused by the exclusion criteria applied inthis study. There is similarity between thestudies in the use of UI programs anddevices. Brandeis et al noted that 39% ofincontinent and 12% of continent residentsreceived scheduled toileting, and that 84% ofincontinent and 22% of continent residentsused pads/briefs.16
In this study, a comparable 10% of continent,and 23% to 42% of incontinent residents(by UI level) had scheduled toileting. Here,only 5% of continent patients usedpads/briefs; but a similar range of 37% to85% of incontinent patients used them(depending on UI level).
As with any database analysis, there arelimitations that should be considered in theinterpretation of the results. This was across-sectional analysis that relied on MDSand other data reported by nursing homecaregivers and caregiver observations. Therewas no indication that caregiver reportingwould have differed by treatment or UI level.The statistically significant results identifiedmay not be clinically meaningful. Large populationswere studied and multiple endpoints were examined; thus, level of significancemay be affected.
There is a high prevalence of UI amongnursing home residents. Residents with UIare more frail than those without UI, and havingthis condition is negatively correlatedwith measures of resident health status (eg,ADLs) and healthcare utilization. A variety ofinterventions are used in this setting to treatUI; however, the use of pharmacologic therapyappears to be quite low. This may becaused by clinician uncertainty regardingwho is the best candidate for drug treatment.Evidence-based approaches, supported bydata specific to the nursing home setting,should be developed to determine whichnursing home residents with UI would benefitmost from pharmacologic therapy. Appropriateuse of interventional strategies thatmay include drug treatment for UI in thenursing home may reduce the substantialpersonal and cost burdens associated withthis condition. This warrants further study inthe nursing home setting to help inform clinicaldecision making regarding UI treatment.
This study was supported by Pfizer Inc.We would also like to acknowledge the assistanceof Eleanor M. Perfetto, PhD, MS,Senior Director, THE WEINBERG GROUPINC, for assistance with manuscript preparation;John Van Vleet and Billy Strunk fortheir support in the study implementation;and Prasun Subedi and ChristopherBlanchette, doctoral students at the Universityof Maryland, Department of PharmaceuticalHealth Services Research, forassistance in assembling the literature.
1. Fantl AJ, Newman DK, Colling J, et al. UrinaryIncontinence in Adults: Acute and Chronic Management.Clinical Practice Guideline Number 2 (1996 Update).AHCPR Publication No. 96-0682: March 1996.
2. Chutka DS, Takahashi PY. Urinary incontinence inthe elderly. Drug treatment options. 1998;56:587-595.
3. Wilson L, Brown JS, Shin GP, Luc KO, Subak LL.Annual direct cost of urinary incontinence. 2001;98:398-406.
Clin Geriatr Med.
4. Tannenbaum C, DuBeau CE. Urinary incontinence inthe nursing home: practical approach to evaluation andmanagement. 2004;20:437-452.
5. Aditya BS, Sharma JC, Allen SC, Vassallo MV.Predictors of a nursing home placement from a nonacutegeriatric hospital. 2003;17:108-113.
6. Hu TW, Wagner TH, Bentkover JD, Leblanc K, ZhouSZ, Hunt T. Costs of urinary incontinence and overactivebladder in the United States: a comparative study.2004;63:461-465.
7. Wagner TH, Hu T. Economic costs of urinary incontinencein 1995. 1998;51:355-361.
J Am Geriatrics
8. Wagner TH, Subak LL. Evaluating an incontinenceintervention in nursing home residents. 2003;51:275-276.
9. Shih YC, Hartzema AG, Tolleson-Rinehart S. Laborcosts associated with incontinence in long-term carefacilities. 2003;62:442-446.
10. National Kidney and Urological Disease InformationClearinghouse. Urinary Incontinence in Women. Availableat: http://kidney.niddk.nih.gov/kudiseases/pubs/uiwomen/index.htm. Accessed December 28, 2004.
Am J Manag Care.
11. Lemack GE. Overactive bladder: optimizing qualityof care. 2001;7:S46-S61.
12. Pharmetrics, data on file. Report to PfizerPharmaceuticals. 2004.
13. Saliba D, Solomon S, Rubenstein L. Feasibility ofquality indicators for the management of geriatric syndromesin nursing home residents. 2004;5:310-319.
J Am Geriatr Soc.
14. Johnson TM, Ouslander JG, Uman GC, Schnelle JF.Urinary incontinence treatment preferences in long-termcare. 2001;49:710-718.
Ostomy Wound Management.
15. Enriquez EL. A nursing analysis of the causes andapproaches for urinary incontinence among elderlywomen in the nursing home. 2004;50:24-43.
J Am Geriatr Soc.
16. Brandeis GH, Baumann MM, Hossain M, et al. Theprevalence of potentially remediable urinary incontinencein frail older people: a study using the minimumdata set. 1997;45:179-184.
17. Morris JN, Fries BE, Mehr DR, et al. MDS CognitivePerformance Scale. 1994;49:M174-M182.
J Am Med Dir Assoc.
18. Ouslander JG, Maloney C, Grasela TH, et al. Implementationof a nursing home urinary incontinence programwith and without tolterodine. 2001;2:207-214.
J Wound Ostomy Continence Nurs.
19. Pinkowksi PS. Urinary incontinence in the long-termcare facility. 1996;23:309-313.
J Am Geriatr
20. Palmer MH, Johnson TM. Quality of incontinencecare in U.S. nursing home: a failing grade. 2003;51:1810-1812.
21. Schnelle JF, Cadogan MP, Yoshii J, et al. The minimumdata set urinary incontinence indicators: do theyreflect differences in processes related to incontinence.2003;41:909-922.
J Am Geriatr Soc.
22. Burgio KL, Locher JL, Goode PS. Combined behavioraland drug therapy for urge incontinence in olderwomen. 2000;48:370-374.
J Adv Nurs.
23. Robinson JP. Managing urinary incontinence in thenursing home: resident's perspectives. 2000;31:68-77.