Urinary Incontinence is an Independent Risk Factor for Nursing Home Placement Among Assisted Living Residents

Abstract

Objectives:

Prior studies suggested that urinary incontinence (UI) may be a risk factor for nursing home (NH) placement among older community-dwelling individuals. Our objectives were to evaluate if UI is an independent risk factor in NH placement among assisted living (AL) residents and assess the impact of UI on NH placement by race/ethnicity.

Design:

This retrospective cohort study is based on the CY2019–2021 Medicare enrollment and claims data.

Setting and Participants:

Medicare beneficiaries residing in AL communities, and those with a new UI diagnosis, between January 1, 2020 and June 30, 2020, were identified. All residents were followed through December 31, 2021 to identify permanent NH placement (length of stay ≥90 days).

Methods:

Individual covariates (age, race/ethnicity, chronic conditions) and AL community factors (bed size, proportion of Medicare-Medicaid dually eligible residents) were included. We estimated logistic regression models with individual and AL random effects and robust standard errors. Interaction effects were estimated to examine differences in NH placement by race/ethnicity.

Results:

Of 247,010 AL residents, 20.45% had UI. Overall, 15.7% of residents were permanently placed in NH during the study period; 14.2% had no UI, and 21.7% had UI. After controlling for confounders, UI significantly (P <0.01) increased the probability of NH placement, a 14% increase relative to the mean. Compared to Whites without UI, Whites with UI had 1.93 percentage point (pp) higher probability of NH placement. While Black and Hispanic residents overall had lower risk of placement, that risk increased significantly for residents with UI; 3.43 pp higher for Blacks and 3.68 pp higher for Hispanics.

Conclusions and Implications:

This first and largest study of UI in AL communities demonstrated that UI is an independent risk factor for permanent NH placement. Education and identifying and treating UI early may allow AL residents to avoid or delay NH placement.

Brief summary:

Early treatment and management of urinary incontinence, an independent risk factor for nursing home (NH) placement among assisted living residents, could potentially delay or prevent NH placement.

INTRODUCTION

Urinary incontinence (UI) is a common and often underdiagnosed condition among older adults.¹ In the US, the prevalence of UI has been reported at over 40% among individuals older than 65 years who live in the community, and as high as 70% among those who are residents of nursing homes.² UI prevalence rates are highest among women, people of advanced age, and those who have cognitive and physical disabilities.^{5, 6} Falls, pressure ulcers, social isolation, depression, and caregiver stress are common adverse consequences of UI.³ UI has been demonstrated to reduce self-reported quality of life, as measured by lower scores for mood, dignity, and meaningful activity among nursing home residents.⁴ In addition, caring for older adults with UI significantly adds to the caregivers’ burden, and increased severity of incontinence is associated with an increased burden of care.⁵ Among community residents, UI has been shown to play a decisive role in shaping their perception of the need to move from their home to the nursing home setting.⁶

While some research has focused on the relationship between UI and nursing home placement,⁷ studies of UI in assisted living communities have been largely absent.⁸ Assisted living communities have evolved in response to older adults’ needs for personal care and a preference to remain in a home-like environment when no longer fully independent.⁹ Today, assisted living communities are the largest residential care setting in the US, accounting for almost 90% of all senior residential care community residents, and with 7 out of 10 people requiring assisted living in their lifetime, this has been the fastest-growing long-term setting.¹⁰ One of the goals and attractions of assisted living is to facilitate residents’ ability to age in place and to delay or prevent nursing home admission.¹¹ However, this has proven to be quite variable and difficult to achieve as both the services, clinical and supportive, provided in this care setting and staffing patterns vary substantially across communities and states.^12,13

UI as a possible risk factor for nursing home placement has been examined in several prior, now relatively old studies. For example, a study by Young et al. showed that the receipt of home services delayed nursing home placement among older community-living individuals, and that the delay was accompanied by significant increases in the prevalence of chronic conditions, including UI.¹⁴ More directly, Nuotio et al., showed that urge incontinence was a significant prognostic indicator of institutionalization among older, community-dwelling men.¹⁵ Several studies done outside of the US also showed that incontinence was a predictor of institutionalization in Japan, New Zealand and Belgium.^16–18 However, the evidence for UI as a predictor of nursing home placement has not been unequivocal. For example, Holroyd-Leduc et al., found that although UI appeared to be a marker of frailty in community-dwelling elderly, it was not an independent risk factor for nursing home admission.¹⁹ Similarly, Friedman et al., studied participants enrolled in Medicare- and Medicaid-capitated All-Inclusive Care for the Elderly (PACE) programs and found UI not to be independently predictive of institutionalization.²⁰ However, PACE is a unique program that is at financial risk for all services including nursing home, and therefore it is more likely to implement interventions to delay or prevent institutional care than may be expected of programs that do not bear financial risk. A systematic literature review, by Luppa et al., concluded that UI is an inconsistent predictor for nursing home admission and questioned the methodical quality of relevant, existing studies.²¹

Most of these existing studies were based on small, convenience data samples and surveys, and are quite dated. To date, no study has examined UI as an independent risk factor for nursing home admission among residents of assisted living communities.

To fill this gap in the literature, we focused on Medicare beneficiaries who resided in assisted living communities in the US in order to: 1) examine whether UI is an independent predictor of nursing home placement for assisted living residents; and 2) assess whether UI effect on nursing home placement differs by race/ethnicity.

METHODS

Data Sources and Study Sample

Following the study’s review and approval by the University of Rochester institutional review board, we obtained data from the CY2019–2021 Medicare Enrollment Database (MED), the Medicare Beneficiary Summary Files (MBSF), the Minimum Data Set (MDS), MedPAR, and outpatient and carrier claims. Using a previously developed methodology, we identified Medicare beneficiaries who were assisted living residents²² in January 1, 2020 through June 30, 2020, and were alive through December 31, 2021. Furthermore, we identified assisted living residents newly diagnosed with a medical diagnosis of UI between January through June of 2020 and followed all residents (those with and without UI diagnosis) through 2021 to identify those permanently placed in a nursing home.

We limited the study cohort to residents aged 55 years and older because they represent the typical assisted living population of Medicare beneficiary residents. Because information on diagnoses for beneficiaries enrolled in the Medicare Advantage (MA) plans are not consistently available in the Medicare claims data, those with any MA enrollment were excluded.

Outcome: Nursing Home Placement

Placement in nursing homes from June 30, 2020 to September 30, 2021 was identified using the MDS and defined as custodial (long-term) if a stay was equal to or longer than 90 days.²³

Key Independent Variables

The key independent variable of interest was a diagnosis of UI (dichotomous), which was identified using ICD-10 codes from the carrier claims, MedPAR, and outpatient claims data. ICD-10 codes are used globally to standardize the classification and documentation of diagnoses, diseases, and health conditions, facilitating healthcare billing, research, and statistical analysis. Assisted living residents with at least one of the following ICD-10 codes were classified as having UI: F980, F981, K5909, K629, R159, L24A2, L258, N318, N319, N3281, N819, N343, N3642, R32, N393, N39490, N394, N3941, N3942, N3943, N3944, N3945, N3946, N3949, N39491, N39492, N39498, N401, N39498, R32, R350, R3915, N9989, N393, O99891, R15, R152, R159, R150, K5900, R194, R198, R3589, R3915, R3981, Z789, Z8719, Z87448, Z87898, Z9189, and Z98890.

Other Covariates

Individual-level covariates included age, gender, race/ethnicity (White, Black, and Hispanic), length of assisted living residence, Medicare-Medicaid dual eligibility, and presence of other chronic conditions (e.g., dementia, diabetes, depression, etc.), all shown to be risk factors for nursing home admission.²⁴ Having been seen by a urologist in calendar years 2020 and 2021 was also included as a covariate, as it may be associated with active UI treatment. Medicare-Medicaid “dual eligible” refers to individuals who qualify for both Medicare (federal insurance) and Medicaid (federal/state entitlement program) benefits, allowing them to receive a broader range of healthcare services at reduced out-of-pocket costs. The information on age, gender, race/ethnicity, and Medicare-Medicaid dual eligibility were determined based on the MBSF. The length of assisted living residence was calculated using the MED. Whether a urologist saw the resident was a binary variable (yes/no) based on the provider specialty codes in outpatient claims, MedPAR, and carrier claims files. The presence of other chronic conditions (dichotomous variable) was derived directly from the MBSF chronic conditions file.

Assisted living community-level covariates such as the proportion of Medicare-Medicaid dually eligible residents and the community size (based on the number of beds) were also included, as indicators of assisted living-level resource availability.⁹ For easier interpretation, following a previously decribed practice,²⁵ we categorized the proportion of Medicare-Medicaid dually eligible residents in an assisted living community as lower than or equal to or greater than 20%, which represents the median of our study cohort. Assisted living community size (number of beds) was categorized at fewer than or equal to or greater than 80 beds, based on the study cohort’s median.²⁵

Statistical Analyses

We first examined bivariate associations (using chi-square and t-tests) for nursing home placement versus none, on all covariates of interest, i.e., UI diagnosis, assisted living residents’ characteristics, and the attributes of assisted living communities where they resided. We then fit a logistic regression model with assisted living community-level random effects, and robust standard errors, to examine the association between UI diagnosis and nursing home placement, controlling for all covariates. Interaction effects were then estimated to examine how the relationship between UI and nursing home placement vary by race/ethnicity. For ease of interpretation, the results from the interaction effects are presented as average marginal effects. Full results of the logistic regression are presented in the supplemental (Table S1).

All statistical analyses were performed using a standard software package (Stata, version. 14.0; Stata Corp).

RESULTS

Of the 247,010 assisted living residents in the study cohort, 20.45% had UI. Overall, 15.7% were permanently placed in a nursing home during the 18-month follow-up study period; 14.2% among those with no UI and 21.7% among those with UI (Table 1). Compared to residents without UI, those with UI were slightly older (83.1 years vs 82.9 years), more likely to be non-Hispanic White (89.7% vs 87.2%), and more likely to be dually eligible for Medicare and Medicaid (28.0% vs 24.3%). Residents with UI were also more likely to experience co-morbid chronic conditions compared to those without UI such as, for example, Alzheimer’s disease and related dementias (35.6% vs 18.4%, respectively), chronic heart failure (33.7% vs 16.9%, respectively), pressure ulcers (24.2% vs 12.1%, respectively), or depressive moods (56.2% vs 30.0%, respectively).

Table 1.

Sample Characteristics by Urinary Incontinence Status

Variables	Does not have UI (n = 196,472)	Has UI (n = 50,538)	Total (N = 247,010)	P value
Nursing home placement	14.2%	21.7%	15.7%	<0.001
AL community factors
Size (number of beds) ≥80	50.5%	53.4%	51.1%	<0.001
Dual eligibility ≥5%*	16.9%	17.7%	17.1%	<0.001
AL resident factors
Age, y, mean (SD)	82.9 (11.2)	83.1 (10.9)	82.9 (11.2)	<0.001
Race/ethnicity
Non-Hispanic White	87.2%	89.7%	87.7%	<0.001
Non-Hispanic Black	5.10%	4.66%	5.01%
Hispanic	3.45%	2.95%	3.35%
Other	4.23%	2.71%	3.92%
Female gender	68.4%	65.0%	67.7%	<0.001
Length of residence, mos, mean (SD)	38.1 (31.7)	36.1 (31.4)	37.7 (31.7)	<0.001
Dual eligibility*	24.3%	28.0%	25.0%	<0.001
Chronic conditions
Chronic kidney disease	20.6%	38.3%	24.2%	<0.001
COPD	17.1%	32.8%	20.3%	<0.001
CHF	16.9%	33.7%	20.3%	<0.001
Diabetes	23.2%	38.3%	26.3%	<0.001
Osteoporosis	17.7%	31.1%	20.4%	<0.001
Mobility impairment	7.06%	13.9%	8.46%	<0.001
ADRD	18.4%	35.6%	21.9%	<0.001
Drug abuse	6.80%	12.8%	8.02%	<0.001
Hip fracture	1.41%	4.73%	2.09%	<0.001
Drinking	6.21%	9.34%	6.85%	<0.001
Obesity	22.4%	35.3%	25.0%	<0.001
Pressure ulcer	12.1%	24.2%	14.6%	<0.001
Bipolar disorder	9.65%	17.5%	11.3%	<0.001
Major depression	36.8%	58.8%	41.3%	<0.001
Depressive moods	30.0%	56.2%	35.3%	<0.001
Anxiety	33.7%	55.5%	38.2%	<0.001

Values are reported as % unless otherwise noted.

^*Indicates AL residents dually eligible for Medicare and Medicaid.

ADRD, Alzheimer’s disease and related dementias; AL, assisted living; COPD, chronic obstructive pulmonary disease; CHF, chronic heart failure; mos, months; y, years.

After controlling for other risk factors (Table 2), UI significantly (P <0.01) increased the probability of nursing home placement by 14% relative to the mean (2.2 percentage point [pp] increase). A visit to a urologist was not associated with nursing home placement.

Table 2.

Marginal Probability of Nursing Home Placement Among Medicare Beneficiaries Residing in Assisted Living Communities: Association with Urinary Incontinence

Covariates	Marginal probability	Standard error	P>z	95% CI low	95% CI high
AL resident factors
Age, y	0.002843	9.32E-05	<0.01	0.002661	0.003026
Race (ref = White)
Black	−0.02259	0.003335	<0.01	−0.02912	−0.01605
Hispanic	−0.02654	0.004055	<0.01	−0.03448	−0.01859
Other	−0.03611	0.004047	<0.01	−0.04404	−0.02818
Urinary incontinence	0.021557	0.001945	<0.01	0.017744	0.02537
Urology visit	0.001228	0.004271	0.774	−0.00714	0.009598
Female gender	0.002485	0.001642	0.13	−0.00073	0.005703
ADRD	0.025252	0.001913	<0.01	0.021503	0.029001
Length of residence, mos	−8.46E-06	8.33E-07	<0.01	−1E-05	−6.83E-06
Dual eligibility*	0.084278	0.002861	<0.01	0.07867	0.089887
Chronic kidney disease	0.01299	0.001752	<0.01	0.009557	0.016424
COPD	0.013101	0.001864	<0.01	0.009448	0.016754
CHF	0.037347	0.001943	<0.01	0.033538	0.041155
Diabetes	0.010807	0.001737	<0.01	0.007403	0.014212
Osteoporosis	0.002957	0.001846	0.109	−0.00066	0.006575
Mobility impairment	0.036697	0.002586	<0.01	0.031628	0.041766
Drug use	0.017394	0.00288	<0.01	0.011749	0.023039
Hip fracture	0.171455	0.006274	<0.01	0.159158	0.183751
Drinking	0.008279	0.002996	0.006	0.002407	0.01415
Obesity	0.014068	0.001768	<0.01	0.010603	0.017533
Pressure ulcer	0.049629	0.002155	<0.01	0.045405	0.053853
Schizophrenia	0.000455	0.002533	0.858	−0.00451	0.00542
Bipolar disorders	0.014689	0.002664	<0.01	0.009469	0.01991
Major depression	0.012451	0.002435	<0.01	0.007679	0.017222
Depressive moods	0.010925	0.002542	<0.01	0.005942	0.015908
Anxiety	0.00252	0.001769	0.154	−0.00095	0.005987
AL community factors
Size (number of beds) ≥80	0.019508	0.002022	<0.01	0.015545	0.023471
Dual eligibility ≥20%*	−0.00935	0.002493	<0.01	−0.01424	−0.00447

^*Indicates AL residents dually eligible for Medicare and Medicaid.

ADRD, Alzheimer’s disease and related dementias; AL, assisted living; COPD, chronic obstructive pulmonary disease; CHF, chronic heart failure; mos, months; y, year.

Furthermore, among White assisted living residents, UI was associated with a 1.93 pp increase in the probability of being placed in a nursing home (Figure 1). While Black and Hispanic residents overall had a lower risk of nursing home placement, that risk increased significantly for those with UI: 3.43 pp higher for Blacks and 3.68 pp higher for Hispanics.

Figure 1.

Average marginal effect of urinary incontinence on nursing home placement by race/ethnicity.

DISCUSSION

This is the first large, national-level study of UI among assisted living residents in the US. We demonstrated that UI is an independent risk factor that significantly increased the probability of nursing home placement (14% increase relative to the mean). Furthermore, our study showed the presence of racial/ethnic differences in nursing home placement with UI being a risk factor.

Our study confirms the findings previously shown among Canadian assisted living residents.²⁶ The Canadian study completed resident assessments for 1086 assisted living residents and interviewed caregivers and administrators to determine predictors of long-term care placement using a multivariable Cox proportional hazard model. UI was identified as a statistically significant predictor in addition to poor social relationships, cognitive and/or functional impairment, health instability, recent falls, and hospitalizations/emergency department visits.

We also examined whether UI differentially affects nursing home placement by race and ethnicity of assisted living residents. We found that while Black and Hispanic residents overall had a lower risk of nursing home placement, that risk increased significantly for those with UI, exceeding the risk of NH placement among comparable White residents. Such disparities have been reported by others as well. In a study conducted in nursing homes located in the southeast of the US, researchers found that the prevalence of UI at admission was 7% higher among Blacks than Whites.²⁷ Bliss et al. analyzed 111,640 admissions to for-profit nursing homes and reported the prevalence of UI at admission to be 38% among Whites, 45% among Hispanics, and 52% among Blacks.²⁸ While 22% of nursing home residents reported to be cured of UI, fewer Hispanics had their incontinence cured and did so at a later time after admission than expected.²⁹

Although assisted living communities provide limited medical and health services, the availability of personal care may be insufficient due to poor staff training and staffing,³⁰ such services often influence residents’ outcomes. For example, studies have shown that hospital admissions and readmissions, emergency room use, and nursing home placement for assisted living residents may be influenced by the services and the environment in which they reside.^24,31–33 UI is a modifiable factor with ever-evolving treatment options for stress and urge incontinence.^34–36 Treatment options implemented in assisted living communities have the potential to change the trajectory of assisted living-to-nursing home placement for many residents with UI, allowing them to age in place.

Although our study is based on a very large cohort (247,010) and is methodologically sound, several limitations need to be acknowledged. First, we were not able to control for UI severity, which may influence nursing home placement as severe UI will place greater demands on assisted living staff time. However, because we used claims data to identify UI, it is possible that cases of less severe UI that were yet not diagnosed were not included, suggesting that those with a UI diagnosis may have a more severe condition. Second, while we tried to control for visits to a urologist that may indicate an opportunity for active UI treatment, we were not able to control for such treatment, the lack of which may influence nursing home placement. Third, Medicare claims capture a medical diagnosis of UI, but individuals who are incontinent but are not diagnosed were not identified. Residents are often embarrassed by UI and may not seek medical help. Thus, the prevalence of UI in this population may be undercounted. Fourth, due to lack of data, Medicare Advantage enrollees were not included, thus our findings may not be generalizable to this growing proportion of the assisted living population. Finally, while we controlled for a host of covariates, as in any observational study it is possible that additional unobserved factors associated with both nursing home placement and UI could bias the findings.

CONCLUSIONS AND IMPLICATIONS

This first study of UI in assisted living residents demonstrated that it is an independent risk factor for permanent nursing home placement. Educating residents, family members, and providers, and identifying and treating UI early in the course of the disease may allow assisted living residents to avoid or delay nursing home placement.

Supplementary Material

MMC!

Table S1. Likelihood of Nursing Home Placement Among Medicare Beneficiaries Residing in Assisted Living Communities: Logistic Regression Analysis