Prevalence of Cognitive Impairment in Home Health Physical Therapy

Abstract

Background:

The prevalence of cognitive impairment in home health physical therapy (HHPT) is unknown. We sought to identify the prevalence of cognitive impairment, including cognitive impairment no dementia (CIND) and dementia, among older adults who used HHPT, and if cognitive impairment prevalence was higher among those with HHPT-relevant characteristics.

Methods:

For our cross-sectional analysis, we identified 963 fee-for-service Medicare beneficiaries with HHPT claims (>85 years old: 28.8%, women: 63.7%, non-Hispanic White: 82.1%) in the 2014 and 2016 waves of the Health and Retirement Study (HRS) and used a validated algorithm to categorize cognitive status as normal, CIND, or dementia. We estimated the population prevalence and calculated age, gender, race/ethnicity adjusted odds ratio (aOR) of CIND and dementia for characteristics relevant to HHPT service delivery including depression, walking difficulty, fall history, incontinence, moderate-vigorous physical activity (MVPA) ≤1x/week, and community-initiated HHPT using multinomial logistic regression.

Results:

The population prevalence of cognitive impairment was 46.4% (CIND: 27.3%, dementia: 19.1%). The prevalence of cognitive impairment was greater among those with depression (46.7% vs 39.5%), difficulty walking across the room (58.9% vs 41.8%), fall history (49.1% vs 42.9%), MVPA ≤1x/week (50.0% vs 38.0%), and community-initiated HHPT (55.2% vs 40.2%). Compared to normal cognitive status, the odds of cognitive impairment were greater for those with MVPA≤1x/week (CIND: aOR=1.57 [95% CI: 1.05-2.33], dementia: aOR=2.55 [95% CI: 1.54 – 4.22]), depression (dementia: aOR=1.99 [95% CI: 1.19 – 3.30]), difficulty walking across the room (dementia: aOR=2.54 [95% CI: 1.40 – 4.60]), fall history (dementia: aOR=1.85 [95% CI: 1.20 – 2.83]), and community-initiated HHPT (dementia: aOR=1.72 (95% CI: 1.13 – 2.61]).

Conclusion:

There is a high prevalence of CIND and dementia in HHPT, and no characteristics had a low prevalence of cognitive impairment. Physical therapists should be ready to identify cognitive impairment and adapt home health service delivery for this vulnerable population of older adults.

Introduction

Older adults commonly use home health services, a Medicare benefit that covers specialized services like nursing care, rehabilitation, and intermittent home health aide care for those who are considered homebound. Of home health services, home health physical therapy (HHPT) use is frequent among older adults, comprising one-third of home health fee-for-service visits, and is most common following acute illness, hospitalization, and community-based functional decline.¹ In an effort to optimize measures that matter, physical therapists in the home health setting develop treatment plans that address complex needs of homebound older adults and their care partners.^2,3 Interventions employed in HHPT include progressive resistive exercises, balance training, patient and caregiver education, home modification recommendations, and other approaches that can be physically and cognitively demanding and influence aging-related outcomes.^4–6

Cognitive impairment is a factor that complicates the delivery of high-quality healthcare and is present among an estimated 44% of Medicare beneficiaries using home health services.^7,8 Cognitive impairment of any severity is common among older adults and confers an elevated risk of injurious falls, hospital readmissions, mortality, and numerous other adverse outcomes when compared to older adults without cognitive impairment.^7,9 Beyond the high prevalence of dementia, there are also a substantial number of older adults who have cognitive impairment that does not impact routine daily functioning or meet diagnostic criteria for dementia, referred to as CIND (CIND: cognitive impairment no dementia) within epidemiologic studies or mild cognitive impairment (MCI) in clinical settings.^7,10 Importantly, the relationships between cognitive impairment and adverse outcomes are underestimated when including only those with dementia and excluding those in the spectrum of cognitive impairment without dementia (eg, CIND).¹¹ These two stages of cognitive impairment also represent two clinically meaningful groups, confer different risks for health outcomes, and would be helpful to study separately.⁷

The identification of cognitive impairment among older adults is a clinically significant imperative.⁷ Achieving this imperative is particularly challenging in the home health setting when communication barriers during transitions of care may preclude ADRD identification upon initiation of home health services. For example, an estimated 60% of older adults did not have pre-existing ADRD documented in the electronic medical record upon home health service delivery initiation.¹² Given these challenges, there is a growing body of research that examines potential indicators of cognitive impairment and decline in the home health setting. Burgdorf et al identified that patients with a home health clinician rating of cognitive impairment have a 6-7 times likelihood of having an ADRD diagnosis within 1 year of home health service delivery when compared to those without a cognitive impairment rating.¹³ Topaz et al also applied natural language processing to electronic medical record data to identify the presence of neuropsychiatric ADRD symptoms among older adults who are using home health services.¹⁴ Despite this evidence, more research is needed to guide earlier identification of cognitive impairment among older adults.

There are a number of potential benefits to earlier identification of cognitive impairment including the use of interventions or clinical care pathways that address symptoms when they are easier to effectively manage, minimize risk factors for cognitive decline and other adverse health outcomes, or optimize patient and care partner quality of life among others.^7,15 In addition to contributing to these potential benefits as an interdisciplinary home health team member, physical therapists’ earlier identification of cognitive impairment could also inform the engagement of care partners and tailoring of rehabilitation treatment plans, including exercise and recommendations,¹⁶ to meet the cognitive capabilities of older adults. We previously estimated that 20% of Medicare Fee-for-service beneficiaries who received outpatient physical therapy services had cognitive impairment,¹⁷ but these findings are not readily generalizable to the specialized setting of HHPT and the prevalence of CIND and dementia among Medicare beneficiaries who use HHPT is largely unknown. Further, characteristics that are both relevant to rehabilitation practice (eg, walking difficulty, fall history) and potential indicators of cognitive impairment are understudied. We need to address these knowledge gaps to inform which older adults should be screened for cognitive impairment, directed towards cognitive impairment clinical care pathways, and potential intervention tailoring strategies for the cognitive abilities of older adults. Therefore, our purpose was to estimate the prevalence of cognitive impairment (CIND and dementia) among Medicare beneficiaries who have HHPT claims. We also sought to identify if the prevalence of cognitive impairment was higher among patient characteristics that are relevant to rehabilitation practice.

Methods

Data sources

We used data from the 2014 and 2016 waves of the Health and Retirement Study (HRS).¹⁸ The HRS is a longitudinal, nationally representative cohort study of adults over 50 years old conducted by the University of Michigan and supported by the National Institute of Aging and Social Security Administration. Briefly, participants are selected using a multistage probability sampling design and surveyed every two years in a variety of topic areas (e.g., cognition, finances, health, functional limitation). In addition, we supplemented the data collected from HRS participants through a Medicare data linkage. Through this data linkage, we identified HRS participants who had claims for at least one HHPT visit.

Participants

We identified 18313 community-dwelling participants 65 years and older in the 2014 and 2016 waves of the HRS. Of those, we identified 16681 records for participants in the 2014 and 2016 waves of HRS who had their Medicare claims linked with HRS data. We then used Revenue Center codes (0420, 0421, 0422, 0423, 0424, 0425, 0426, 0427, 0428, 0429) in the CMS home health file to identify participants with at least one fee-for-service HHPT visit claim in the year following the date of HRS data collection. Our final cohort of community-dwelling older adults who had a fee-for-service HHPT visit claim included 963 participant records (2014: 496 records, 2016: 467 records) for 897 unique participants.

Measures

Primary Measure:

Cognitive status was our primary measure of interest and was classified using the previously validated Langa-Weir algorithm into normal, CIND, and dementia groups.^19–21 The algorithm uses the HRS cognitive scale (score range: 0 – 27), which consists of tests of immediate and delayed recall of ten common nouns, serial subtractions by seven, and a backward count task from 20.^19–22 Briefly, participants with a cognitive scale score of ≤6 points were classified as having dementia, or severe cognitive impairment that influences everyday function. Participants with a cognitive scale score of 7-11 were classified as having CIND, a broad category that includes cognitive impairment that includes MCI and other forms of cognitive impairment with independent completion of everyday tasks, yet difficulty with new or cognitively demanding tasks.¹⁰ Participants with a cognitive scale score of ≥12 were classified as normal. If an HRS participant was unable to answer cognitive assessment questions, cognitive status was determined using data from proxy respondents and HRS interviewer. Proxy respondents provided data regarding a direct assessment of memory ranging from excellent to poor (score range: 0 - 4) and limitations in five instrumental activities of daily living (managing money, taking medication, preparing hot meals, using phones, doing groceries; score range: 0 – 5). HRS interviewers also rated difficulty completing the data collection interview because of cognitive impairment (score range: 0 – 2). Combining data from the proxy respondent and HRS interviewer, participants were classified as having normal cognition (0-2), CIND (3-5) or dementia (6-11).^19–21

Additional Measures of Interest:

We selected the following measures because they represent factors relevant to physical therapist practice and may be indicators for cognitive impairment among older adults: depression, walking difficulty, fall history, incontinence, moderate-vigorous physical activity (MVPA) frequency, and community-initiated HHPT.^6,23–25

The presence of depression was indicated by a score of >2 using the short Center for Epidemiological Studies Depression (CES-D) scale.²⁶ We classified severity of walking difficulty based on participant responses to three yes/no questions, “Because of a health problem do you have any difficulty with walking [several blocks, one block, across the room]?”. Fall history was identified using a participant response of “yes” to a single survey question about fall history in the past 2-years. We defined fall history as one or more falls because older adults who experience a fall may benefit from interventions including environmental modification, exercise, balance training, and referral to physical therapist services.²⁷ Participants were categorized as having urinary incontinence if they responded “yes” when asked “during the last 12 months, have you lost any amount of urine beyond your control?”.²⁸ Urinary incontinence is prevalent among older adults and non-pharmacologic interventions employed by physical therapists can optimize patient-centered urinary incontinence outcomes,^28–30 thus was included within our analysis. We classified self-reported MVPA as ≤1x/week using two HRS survey questions regarding frequency of moderate and vigorous intensity physical activity.³¹ Finally, participants with an inpatient discharge with 14 days of HHPT initiation in the CMS MedPAR were categorized as having inpatient-initiated HHPT, where those without were categorized as having community-initiated HHPT.

We also used age, gender, race/ethnicity, marital status, high school education attainment, and ≥2 self-reported health conditions (high blood pressure, diabetes, cancer, lung disease, heart disease, stroke, or arthritis) as participant descriptive covariates.

Analysis

We first estimated survey-weighted population prevalence of cognitive impairment (CIND, dementia), additional measures of interest, and descriptive covariates for Medicare beneficiaries with HHPT claims. We then estimated the population prevalence of cognitive impairment (CIND, dementia) by depression, walking difficulty, fall history, incontinence, physical activity frequency, and community-initiated HHPT variables. We also estimated the prevalence of cognitive impairment by descriptive covariates. Finally, we calculated age, gender, race/ethnicity adjusted odds ratio (95%) of CIND or dementia relative to normal cognitive function by our additional measures of interest using multinomial logistic regression. All statistical analyses were conducted using Stata/SE version 16.1 (StataCorp, College Station, TX, USA) and included HRS survey weights to account for the complex sampling design, differential probability of selection in HRS, and clustering of participants who contributed data to both analytic waves (N=66).

Results

Among community-dwelling older adults with Medicare fee-for-service claims for home health physical therapy, 28.8% were >85 years old, 63.7% were women, and 82.1% were non-Hispanic White (Table 1). The survey-weighted population prevalence estimate of cognitive impairment was 46.4% (95% CI: 42.7% – 50.2%), where the prevalence of CIND was 27.3% (95% CI: 24.1% – 30.8%) and dementia was 19.1% (95% CI: 16.4% – 22.1%). The prevalence of cognitive impairment was greatest among those with depression (46.7% vs no depression: 39.5%), difficulty walking across the room (58.9% vs no difficulty walking: 41.8%), history of falls (49.1% vs no history of falls: 42.9%), MVPA ≤1x/week (50.0% vs MVPA >1x/week: 38.0%), and community-initiated HHPT (55.2% vs inpatient-initiated HHPT: 40.2%; Figure 1). Additionally, population prevalence estimates for cognitive impairment were higher among those who were older than 85 years compared to those who were younger than 75 years (64.0% vs 29.4%), Black compared to non-Hispanic White (67.1% vs 42.8%), not married/partnered compared to those who were (53.0% vs 38.8%), less than high school education compared to those with at least high school education (75.4% vs 38.2%), and those with poor self-rated health compared to those with good self-rated health (58.1% vs 38.2%). There was a similar prevalence of cognitive impairment among those with and without urinary incontinence (45.7% vs 46.7%).

Table 1.

Characteristics of Medicare beneficiaries with Home-Health Physical Therapy (n=963)

Characteristics		Survey Weighed Prevalence (%)
Age	≤75	33.6
	75-85	37.7
	>85	28.8
Gender	Man	36.3
	Woman	63.7
Race/Ethnicity	Non-Hispanic White	82.1
	Black	10.0
Married/Partnered	No	52.4
	Yes	47.6
Less than High School Education	No	77.9
	Yes	22.1
Self-Reported Health Conditions	1-2 Conditions	35.3
	>2 Conditions	64.7
Self-Rated Health	Good or better	45.4
	Fair	34.5
	Poor	20.2
Depression	No	67.7
	Yes	32.3
Walking difficulty	None	27.6
	Several Blocks	19.4
	One Block	27.9
	Across Room	25.1
Fall History	No	43.6
	Yes	56.4
Urinary Incontinence	No	52.0
	Yes	48.0
Moderate-Vigorous Physical Activity	>1x/week	30.1
	≤1x/week	69.9
HHPT Initiation	Community	41.2
	Inpatient	58.8

Center for Medicare and Medicaid Services (CMS) guidelines prohibit reporting of cells with 25 participants or less, therefore we have omitted values for those with Hispanic and Other race/ethnicity and report combined prevalence for those with good, very good, and excellent self-rated health. HHPT: Home-Health Physical Therapy

Figure 1.

Survey-weighted population prevalence of CIND (cognitive impairment, no dementia) and dementia across patient characteristics.

MVPA: Moderate-Vigorous Physical Activity, HHPT: Home-Health Physical Therapy.

* Indicates CIND and dementia prevalence varies by variable (P < 0.05).

Compared to older adults with claims for HHPT who have normal cognitive function, the age, gender, race/ethnicity adjusted odds of CIND was 1.57 (95% CI: 1.05-2.33) times higher for those who report MVPA ≤1x/week compared to those who report MVPA >1x/week (Table 2). The adjusted odds of dementia compared to those with normal cognitive function was 1.99 (95% CI: 1.19 – 3.30) times higher for those with vs without depression, 2.54 (95% CI: 1.40 – 4.60) times higher for those with difficulty walking across the room vs no difficulty walking, 1.85 (95% CI: 1.20 – 2.83) for those with vs without history of falls, 2.55 (95% CI: 1.54 – 4.22) for those with MVPA≤1x/week vs MVPA >1x/week, and 1.72 (95% CI: 1.13 – 2.61) for those with community-initiated vs inpatient-initiated HHPT.

Table 2.

Adjusted odds ratio (aOR) of CIND(cognitive impairment, no dementia) and dementia by measures of interest

		Normal	CIND aOR† (95% CI)	Dementia aOR† (95% CI)
Depression*	No	Reference
	Yes		1.08 (0.71 - 1.65)	1.99 (1.19 - 3.30)
Walking Difficulty*	None	Reference
	Several Blocks		0.93 (0.53 - 1.63)	0.94 (0.49 - 1.82)
	One Block		1.02 (0.62 - 1.69)	1.29 (0.70 - 2.38)
	Across Room		1.38 (0.84 - 2.26)	2.54 (1.40 - 4.60)
Fall History*	No	Reference
	Yes		1.09 (0.76 - 1.56)	1.85 (1.20 - 2.83)
Incontinence*	No	Reference
	Yes		0.84 (0.58 - 1.21)	1.24 (0.82 - 1.87)
MVPA ≤1x/week*	No	Reference
	Yes		1.57 (1.05 - 2.33)	2.55 (1.54 - 4.22)
HHPT Referral*	Inpatient	Reference
	Community		1.27 (0.87 - 1.84)	1.72 (1.13 - 2.61)

^†Age, gender, race/ethnicity adjusted multinomial logistic regression,

^*model significant at P < 0.001

MVPA: Moderate-Vigorous Physical Activity, HHPT: Home-Health Physical Therapy

Discussion

We identified that nearly half of fee-for-service Medicare beneficiaries who use HHPT have cognitive impairment. While we also found relationships of cognitive impairment across multiple measures of interest, there were no patient characteristics that had a low prevalence of cognitive impairment. The ubiquity of cognitive impairment among those receiving HHPT has important implications for service delivery.

Our cognitive impairment prevalence estimate of 46.4% is similar to prior work focused on Medicare beneficiaries who used any home health service and were classified as having cognitive impairment when assessed by a clinician using a single OASIS item.⁸ We extended the findings of this prior work by using a validated measure of cognitive status within the HRS to determine the prevalence of cognitive impairment along the spectrum of CIND and dementia within a sub-group of beneficiaries that used HHPT. Specifically, we estimated that 25% of Medicare beneficiaries who receive HHPT have cognitive impairment not severe enough to be dementia and nearly 20% of beneficiaries had dementia.

We also identified that CIND and dementia are more prevalent among older adults with characteristics that are commonly encountered or addressed by physical therapists in the home health setting. Specifically, the presence of depression, walking difficulty across the room, history of falls, MVPA≤1x/week, and community-initiated HHPT translated to a 1.7-2.5 times higher odds of dementia compared to those without these characteristics. Further, those who had MVPA≤1x/week had a 1.5 times higher odds of having CIND compared to those who did not. Prior work has identified that depression and community-initiated home health services are potential indicators of cognitive impairment.^13,14,17 We expand on these findings by identifying physical function indicators of cognitive impairment, and physical inactivity may be an indicator of CIND in HHPT. Knowledge of these physical function indicators of cognitive impairment, especially among those with CIND, are needed to strengthen our clinical ability to identify cognitive impairment earlier in the disease process.

We used population-based data merged with Medicare claims to quantify the presence and relationships of cognitive impairment with HHPT characteristics. The results of the current study highlight differences between those who receive home health versus outpatient physical therapy. In our prior study among Medicare beneficiaries with outpatient physical therapy claims,¹⁷ we estimated that 13% were over the age of 85 years, 23% had depression, 10% had difficulty walking across the room, 49% had a fall history, and 46% reported MVPA≤1x/week. Relative to our prior findings, older adults with HHPT claims were older and had a higher prevalence of depression (32%), walking difficulty across the room (25%), fall history (56%), and MVPA ≤1x/week (70%). The prevalence of cognitive impairment was higher among those in HHPT (CIND: 27%, dementia: 19%) compared to outpatient (CIND: 15%, dementia: 5%). Finally, our estimates of cognitive impairment prevalence by variables of interest ranged from 38-59% in home health physical therapy, which is more than twice the 14-30% prevalence in the outpatient setting. Our findings suggest that older adults who receive HHPT services, when compared to the outpatient setting, are more likely to have severe limitations and cognitive impairment, regardless of potential subgroup characteristics. Therefore, interventions, service delivery, and policy should be designed to address and accommodate for the complex needs of older adults who use HHPT services.

The home health team’s interdisciplinary use of cognitive screening tools provides additional opportunities to identify cognitive impairment and minimize the risk of pre-existing cognitive impairment going unnoticed through the continuum of care. Physical therapists can contribute to identifying cognitive impairment because they are expected to be competent in selecting, administering, and interpreting valid and reliable tools to be used with older adults, including cognitive screening tools, and determine the need for referral.³² Physical therapists can also develop plans of care for older adults with complex medical profiles (eg, older adults with dementia),³² but commonly report numerous barriers to delivering quality care for older adults with cognitive impairment.^33,34 These barriers, including complex systems, knowledge, confidence, or comfort, represent opportunities for HHPT service delivery improvement.

Identification and management of cognitive is complex,^12–14 and addressing these knowledge gaps and implementation barriers could improve health outcomes among older adults. For example, targeted training to address knowledge gaps and implementation challenges (eg, time, confidence) surrounding the use and interpretation of findings from the Brief Interview of Mental Status (BIMS) could be beneficial because it is a newly implemented cognitive screen within the home health Outcome Assessment Information Set (OASIS-E). Further, findings from the BIMS could potentially guide physical therapist tailoring of interventions to meet an older adult’s cognitive capabilities and improve referral to specialized care pathways for cognitively impaired older adults and their care partners. Further research is needed to examine potential disparities of HHPT referrals and outcomes among those with cognitive impairment, and strategies to minimize these disparities among older adults.

Limitations

Although there are strengths in using data from a large, population-based study of older adults linked with Medicare claims, our study was cross-sectional in nature and no inference of causality can be made between cognitive status and our additional measures of interest. There may also be differences among those who consented to the HRS and Medicare claims linkage, have fee-for-service plan, or are dual-eligible (Medicare and Medicaid), thus generalizability could be limited. Additionally, future research is needed to further investigate variables that we did not include in our analysis (eg, number of sessions or episodes).

Conclusions

There is a high prevalence of cognitive impairment, including those with and without dementia, in HHPT. While the prevalence of cognitive impairment was higher among older adults with depression, difficulty walking, fall history, physical inactivity, and community-initiated HHPT, there were no patient characteristics with a low prevalence of cognitive impairment. Physical therapists should be ready to identify cognitive impairment and adapt home health service delivery for this vulnerable population of older adults.

Key Points:

• Nearly half of older adults who use home health physical therapy have cognitive impairment.

• While the prevalence of cognitive impairment varies by characteristics that are relevant to home health physical therapy service delivery (eg, depression, fall history), all subgroups had a cognitive impairment prevalence of at least 38%.

• Further research is needed to develop and test the effectiveness of strategies that address the complex needs of cognitively impaired older adults in the home health physical therapy setting.

Why does this paper matter?

The high prevalence of cognitive impairment across factors relevant to physical therapist practice suggests that physical therapists should be ready to identify cognitive impairment and adapt home health service delivery for the needs of this vulnerable population of older adults.