Family Caregiver Availability and Capacity Associated With Home Health Service Types for Patients With Dementia

Abstract

Objectives:

To investigate the relationship between unmet needs for family caregiver support, defined as lack of caregiver availability or capacity, and home health care (HH) service types delivered to patients with dementia.

Design:

Cross-sectional study of national claims and assessment data.

Setting and Participants:

325,148 older adults (aged ≥65 years) with diagnosed dementia receiving Medicare-funded HH in 2018.

Methods:

Service types were measured from Medicare claims and refers to patient receipt of each of 6 service types covered during HH: nursing, physical therapy, occupational therapy, speech therapy, social work, and personal care aide. We fit multivariable logistic regression models estimating the odds of receiving each service type as a function of unmet caregiving needs and patient and HH agency characteristics, while clustering at the HH agency level.

Results:

In adjusted models, lack of caregiver availability to help with medical, functional, or instrumental tasks was associated with higher odds of receiving skilled nursing [adjusted odds ratio (aOR) 1.63, 95% CI 1.47–1.81], aide (aOR 1.61, 95% CI 1.54–1.68), and social work (aOR 2.71, 95% CI 2.56–2.86), respectively. Lack of caregiver capacity to help with medical, functional, or instrumental tasks was associated with higher odds of receiving skilled nursing (aOR 1.29, 95% CI1.20–1.39), physical therapy (aOR 1.74, 95% CI1.68–1.81), and social work (aOR 1.24, 95% CI 1.17–1.28), respectively.

Conclusions and Implications:

We observed significant associations between unmet needs for caregiving support and HH service delivery for patients with dementia. Expanding HH quality measurement and risk adjustment data elements to include measures of caregiver availability and capacity are necessary steps toward supporting HH providers’ efforts to meet patients’ needs regardless of their caregiving context.

Through the home health (HH) benefit, homebound Medicare beneficiaries with a temporary need for skilled care can receive services delivered in their home. More than one-third (36%) of HH patients have diagnosed dementia.¹ Available evidence indicates that care delivery and outcomes in HH differ meaningfully by dementia status; for example, patients with dementia have longer lengths of stay and are more likely to be hospitalized during HH.^2–6 However, research identifying best practices and determinants of clinical outcomes for HH patients with dementia is sparse compared with research focused on institutional settings.⁵

HH patients with dementia rely heavily on family and unpaid caregiver (hereafter, “caregiver”) assistance with self-care, medical management, and care coordination tasks, including during the HH episode.⁷ Patients may experience unmet needs for caregiver assistance if they require help in a given area, but there is a lack of availability (no caregiver is present) or a lack of capacity (the caregiver needs training to safely and effectively assist).^8–11 HH clinicians have stated that unmet needs for caregiver support impact care planning, including decisions about which service types to deliver.¹²

HH covers 6 services: skilled nursing, physical therapy, occupational therapy, speech therapy, social work, and personal care aide services.¹³ HH clinicians, particularly the registered nurse (RN) or physical therapist (PT) leading the episode, have broad latitude in requesting physician orders for specific service types in response to patient clinical presentation and needs.¹⁴ Service types are a major determinant of costs and profits for the HH agency and shape how a patient experiences HH.¹⁵

A prior study found a relationship between caregiver training needs and visit intensity during HH, but this research did not specifically examine patients with dementia and had a relatively small, survey-based sample.¹⁶ There is no national evidence regarding how unmet caregiving needs may affect HH service types among patients with dementia. We leverage national, linked Medicare HH claims and assessment data to assess whether and how unmet needs for caregiving support, defined by lack of caregiver availability or capacity, are associated with HH service types. Findings are germane to ongoing efforts to develop targeted HH pathways for patients with dementia and to design reimbursement and coverage policies that facilitate post-acute care providers’ ability to meet the unique needs of patients with dementia.^5,10,17–19

Methods

Data Set and Sample

We linked 2018 Medicare HH claims, Outcomes and Assessment Information Set (OASIS), Medicare Beneficiary Summary File (MBSF) base and chronic conditions supplement files, Provider of Services, and Home Health Care Compare data. Medicare HH claims included dates and types (eg, physical therapy) of HH services. OASIS is the standardized HH patient assessment, with information on clinical and functional status and caregiver involvement. MBSF included information on beneficiaries’ sociodemographic characteristics and diagnoses (including dementia). The Provider of Services file captured descriptive information about Medicare-certified HH agencies, and the Home Health Care Compare file reported Centers for Medicare & Medicaid Services (CMS) quality star ratings for HH agencies.

We included older adults (aged ≥65 years) enrolled in Traditional Medicare for whom we observed an HH admission and discharge in 2018 and who had diagnosed dementia (as defined below). We excluded those residing in congregate living at the time of HH, who died during the HH episode, or who had a Low-Utilization Payment Adjustment or denial of payment for HH. We excluded those with a length of stay of less than 28 days, as they may have less opportunity to receive services because of delays in orders and/or visit scheduling (28 days is the 25th-percentile length of stay in our sample). The final analytic sample included 325,148 HH patients with dementia (see Appendix Figure A1 for details regarding application of exclusion criteria). We examine only the index HH certification period (ie, the first 60 days of receiving HH care) for each individual.

Measures

Dementia diagnosis

We determined dementia diagnosis from the MBSF chronic conditions supplement. Individuals were identified as having dementia if they received a claim with a qualifying International Classification of Diseases, Tenth Revision (ICD-10) code over a 3-year lookback period for Alzheimer disease or senile dementia.²⁰

Unmet caregiving needs

We measured unmet caregiving needs from clinician (registered nurse or physical therapist) reports in the start of care OASIS assessment. Among patients identified as needing caregiver help with a specific task category, clinicians note whether no caregiver is available (lack of availability) or the caregiver needs training before she or he can safely assist (lack of capacity). (Details regarding this assessment item and measure construction are included in Appendix section A).

We created binary indicators of lack of availability or lack of capacity by type of caregiving activity: functional, medical, or instrumental help. Functional help refers to assistance with self-care (eg, performing therapy exercises) and activities of daily living (ADL, eg, dressing oneself). Medical help refers to assistance with managing medications or medical equipment (eg, an oxygen machine). Instrumental help refers to assistance with coordinating care or supervision to ensure safety. Construction of these categories is described in greater detail elsewhere.¹¹

Service types

We measured service types from HH claims. Claims data include Healthcare Common Procedure Coding System (HCPCS) codes²¹ indicating the type of care provided at each HH visit. We created binary indicators of whether a patient received 1 or more visits from each of the 6 covered HH service types: skilled nursing, physical therapy, occupational therapy, speech therapy, social work, and personal care aide. We considered visits delivered between admission date and either discharge date or 60 days postadmission (HH patients must be recertified every 60 days to continue receiving care).

Covariates

Patient sociodemographic characteristics include age, race or ethnicity, female sex, and Medicaid enrollment. Clinical status indicators include post-acute status (hospital or skilled nursing facility discharge within 14 days of HH admission), 2 or more ADL impairments prior to home health, depression, presence of a wound, and overall clinical severity. Depression was determined via a score of 3 or higher on the Patient Health Questionnaire—2²² and overall clinical severity drawn from Home Health Resource Group codes.²³ HH agency characteristics included nonprofit ownership, rural service area, affiliation with an acute care hospital, and CMS Star Quality Rating tertile.

Statistical Analyses

We fit multivariable logistic regression models estimating the odds of receiving any visits from each of the HH service types as a function of unmet caregiving needs due to lack of availability or lack of capacity, overall and by type of caregiving assistance needed. Models included cluster robust SEs at the HH agency level and adjusted for patient sociodemographic characteristics and clinical status indicators, and HH agency characteristics. Covariates were chosen to reflect factors associated with unmet caregiving needs in prior work¹¹ and agency-level factors likely to impact service provision. As a sensitivity analysis, we replicated all models while including those with a length of stay of less than 28 days. Analyses were performed using Stata 17 (StataCorp LLC, College Station, TX) and approved by the VNS Health Institutional Review Board.

Results

Sample Characteristics

Among 325,148 HH patients with dementia, 10.9% experienced an unmet need for caregiver assistance due to lack of caregiver availability and 72.2% experienced an unmet need for caregiver assistance due to lack of capacity (Table 1). Most patients received nursing (77.1%) and physical therapy (87.1%) services, whereas smaller proportions received occupational therapy (48.3%), speech therapy (11.9%), social work (16.3%), or personal care aide (18.5%) services. Most (62.7%) were female and non-Hispanic White (76.5%).

Table 1.

Characteristics of Medicare Beneficiaries With Dementia Receiving Home Health Care (N = 325,148)^*

Characteristic	%	n
Unmet needs for caregiving support
Lack of caregiver availability, by type of help needed
Any help	10.9	35,394
Functional help	9.4	30,594
Medical help	4.7	15,336
Instrumental help	4.9	15,863
Lack of caregiver capacity, by type of help needed
Any help	72.2	234,875
Functional help	67.8	220,578
Medical help	58.3	189,648
Instrumental help	46.7	151,703
Home health service mix
Receipt of services, by service type†
Nursing	77.1	250,680
Physical therapy	87.1	283,087
Occupational therapy	48.3	157,065
Speech therapy	11.9	38,728
Social work	16.3	52,900
Personal care aide	18.5	60,059
Patient characteristics
Cognitive symptom severity
Low	25.9	84,304
Moderate	41.3	134,251
High	32.8	106,593
Age, y (quartiles)
65–77	26.5	86,304
78–83	23.9	77,685
84–88	27.2	88,505
≥89	22.4	72,837
Female sex	62.7	203,777
Race or ethnicity
Non-Hispanic White	76.5	248,567
Non-Hispanic Black	11.7	38,164
Hispanic	7.2	23,523
Other	4.6	14,894
Medicaid enrolled	19.3	62,742
Post-acute	59.9	194,834
Functional impairment‡ (tertiles)
Lowest	32.1	104,371
Middle	34.6	112,551
Highest	33.3	108,226
Depressed	3.8	10,922
Wound	29.4	95,612
Overall clinical severity
Low	19.4	63,173
Moderate	30.2	98,030
High	50.4	163,893
Home health agency characteristics
Nonprofit ownership	31.0	100,782
Rural service area	12.7	41,336
Affiliated with acute care hospital	9.6	31,111
CMS Star Quality Rating (tertiles)
Lowest	31.6	101,569
Middle	25.9	83,209
Highest	42.5	136,349

^*Data from linked 2018 Medicare home health claims, Outcomes and Assessment Information Set, Medicare Beneficiary Summary File, Provider of Services files, and Medicare Care Compare data for Traditional Medicare beneficiaries with diagnosed dementia who received home health in 2018 and had a length of stay of at least 28 days.

^†Receipt of 1 or more visits to deliver service type during the first 30 days of index home health episode.

^‡Measured as composite ADL impairment score, as defined in the Measures section.

Association Between Caregiver Availability and HH Service Types

In adjusted models predicting service types, lack of caregiver availability to provide medical assistance was most strongly associated with greater odds of receiving skilled nursing [adjusted odds ratio (aOR1.63, 95% CI 1.47–1.81] and lower odds of receiving physical therapy (aOR 0.64, 95% CI 0.61–0.68) (Figure 1). Lack of availability to provide functional assistance was most strongly associated with receiving personal care aide services (aOR 1.61, 95% CI 1.54–1.68). Finally, lack of availability to provide instrumental assistance was most strongly associated with receiving medical social work (aOR 2.71, 95% CI 2.56–2.86).

Fig. 1.

Adjusted odds of receiving specific home health service types, comparing patients with vs without an available caregiver (N = 325,148). Results of multivariable logistic regression models adjusting for patient sociodemographic characteristics and clinical status indicators, and home health agency characteristics, with cluster robust SEs at the home health agency level.

Association Between Caregiver Capacity and HH Service Types

In adjusted models predicting service types, lack of caregiver capacity to provide functional assistance was most strongly associated with greater odds of receiving physical therapy (aOR 1.74, 95% CI 1.68–1.81) (Figure 2). Lack of capacity to provide medical assistance was most strongly associated with greater odds of receiving skilled nursing (aOR 1.29, 95% CI 1.20–1.39). Finally, lack of capacity to provide instrumental assistance was most strongly associated with receiving medical social work (aOR 1.24, 95% CI 1.17–1.248). (Results of all models are included in Appendix, section B).

Fig. 2.

Adjusted odds of receiving home health service types, comparing patients whose caregiver does vs does not need training (N = 325,148). Results of multivariable logistic regression models adjusting for patient sociodemographic characteristics and clinical status indicators, and home health agency characteristics, with cluster robust SEs at the home health agency level.

Sensitivity Analysis

Including individuals with a length of stay less than 28 days did not produce meaningful shifts in the direction or magnitude of observed relationships. See Appendix Table C1 for these results.

Discussion

Among a national sample of HH patients with dementia, unmet caregiving needs were significantly associated with HH service types. For example, patients with an unmet need for caregiver help with medical tasks were more likely to receive skilled nursing, and those with an unmet need for help with instrumental tasks were more likely to receive social work. Observed relationships between unmet caregiving needs and HH service types indicate that HH care delivery varies with caregiving context and needs for patients with dementia. This aligns with prior studies linking a lack of caregiver capacity to increased HH visit volume.^16,18 Our findings extend previous work by considering service types as an outcome, examining caregiver availability in addition to capacity, and focusing on the high-need patient subpopulation of those with dementia.

Caregiving context and needs are largely ignored in the current HH quality reporting and risk adjustment landscape: no caregiving factors are incorporated in determining performance/payment under the Patient-Driven Groupings Model, Home Health Value Based Purchasing, or CMS Quality Star Ratings.^24–26 Expanding quality measurement and risk adjustment data elements to include measures of caregiver availability and capacity would be vital first steps toward supporting HH providers’ efforts to meet patients’ needs regardless of their caregiving context. Additionally, shifts in the HH payment environment are likely to restrain HH agencies’ flexibility in providing additional services in response to unmet caregiving needs. These include 2020 revisions to the Traditional Medicare HH payment system, which shorten the payment period, and growth in Medicare Advantage (MA) enrollment, with attendant utilization control strategies such as prior authorization, visit limitations, and cost-sharing.^26,27

HH providers have an important role in supporting community-living older adults with dementia. HH services, particularly social work, can help identify and ameliorate vulnerabilities and stressors contributing to adverse outcomes and caregiver strain.²⁸ However, addressing unmet caregiving needs among this population will require intervention beyond the scope of the current Medicare HH benefit. There is a pressing need for coherent strategies to provide comprehensive, coordinated care for community-living older adults with dementia and for these programs to adopt a standardized approach to assessing and supporting family caregivers.

The recently launched CMS demonstration Guiding an Improved Dementia Experience (GUIDE) Model holds promise in this arena. Providers who choose to participate in GUIDE are paid to manage and deliver coordinated services for community-living beneficiaries with dementia, including a limited set of caregiver supports.²⁹ As implementation proceeds, it will be crucial to monitor impacts on dementia caregiver capacity and well-being, in addition to effects on care recipient outcomes. If GUIDE proves effective in improving caregiver efficacy and reducing strain, it will be important to incorporate key aspects of this pilot model into settings that are available to the broader Medicare population, such as HH.

This is an observational study and therefore does not support causal inferences around the relationships observed. Notable limitations to this work include exclusion of Medicare Advantage enrollees (because of a lack of available claims data), inability to comment on the underlying reasons for a lack of caregiver availability or capacity because these are not captured in the OASIS, and a lack of information on the specific care activities performed by each service type for an individual patient. Given reliance on clinician reports to determine caregiver capacity and availability, there may be overlap between these concepts if clinicians interpret a caregiver being unprepared to provide necessary assistance as being “unavailable.” The 2 concepts may be interrelated. Qualitative studies with HH providers could help further delineate between issues of capacity and availability, and elucidate the underlying reasons for the observed associations. Strengths include a large, national sample, rich data on underlying characteristics of individual patients and HH agencies, and ability to examine specific categories of caregiving assistance and types of HH services.

Conclusion

The types of services delivered during an HH episode vary with unmet caregiving needs, with higher provision of services that reflect and address needs that caregivers are unable to fully meet. Expanding HH quality measurement and risk adjustment data elements to include measures of caregiver availability and capacity are necessary steps toward supporting HH providers’ efforts to meet patients’ needs regardless of their caregiving context.

Supplementary Material

Supplementary Data

Supplementary data related to this article can be found online at https://doi.org/10.1016/j.jamda.2025.105744.