Home health utilization in the Veterans Health Administration: Are there rural and urban differences?

Abstract

Purpose

Growing numbers of older adults need home health care, yhese services may be more difficult to access for rural Veterans, who represent one‐third of Veterans Health Administration (VA) enrollees. Our objective was to examine whether home health use differs within VA based on rurality.

Methods

We examined national VA administrative data for 2019–2021 (January 1, 2019 to December 31, 2021) among Veterans ages ≥65 years. Using descriptive and multivariable analyses, we assessed whether rural versus urban Veterans differed in (1) the likelihood of using any home health and (2) for those who received ≥1 visit, number of visits received.

Results

Among home health users (n = 107,229, 33.1% rural), rural and urban Veterans were similar in age (77.0 vs. 77.2 years). Rural Veterans were less likely to be highly frail (38.9% rural vs. 40.4% urban) or diagnosed with dementia (13.5% vs. 17.6%). After adjusting for Veterans’ characteristics, rural Veterans were more likely to receive any home health (odds ratio: 1.10; 95% confidence interval [CI]: 1.07, 1.13). Among Veterans who received ≥1 home health visit, rurality was associated with considerably fewer expected visits (incident rate ratio: 0.70; 95% CI: 0.68, 0.72).

Conclusions

Although rural Veterans were more likely than urban Veterans to receive any home health services, they received considerably fewer home health visits. This difference may represent an access issue for rural Veterans. Future research is needed to identify reasons for these differences and develop strategies to ensure rural Veterans’ care needs are equitability addressed.

INTRODUCTION

Of the 2.7 million rural Veterans who receive care from the Veterans Health Administration (VA), 55% are ages 65 or older. ¹ Older Veterans are more likely to be diagnosed with chronic conditions that require frequent, ongoing care. ¹ , ² Home health services can improve outcomes for those recovering from surgery or serious illness ³ and support older adults who need long‐term services to remain independent at home. ⁴ Yet, accessing home health has become increasingly difficult due to growing demand, workforce shortages, and challenges to service delivery in rural areas, including fewer agencies and transportation barriers. ⁵ , ⁶ , ⁷ Inequitable home health access may result in rural Veterans having greater unmet care needs or relying more heavily on facility‐based care than urban Veterans.

Supportive care needs are met partly through home‐based primary care, a longitudinal program providing interdisciplinary care to some of the VA's most clinically complex Veterans. ⁸ Other skilled home health services in the VA (e.g., nursing, rehabilitation—hereafter “home health”) are provided by VA‐contracted community agencies, which are state‐licensed or certified by the Centers for Medicare and Medicaid Services. To receive home health through the VA, Veterans must demonstrate clinical need and be referred by a VA physician. ⁷ Unlike Medicare, the VA does not require a Veteran to be “homebound” to receive home health. Further, VA's home health services can be delivered long‐term, whereas Medicare typically only covers short‐term or intermittent care. ⁹ , ¹⁰

Medicare studies conducted before the COVID‐19 pandemic found disparities in home health use among rural versus urban beneficiaries (e.g., access to rehabilitation specialists). ¹¹ , ¹² , ¹³ Less is known about rurality‐based differences within the VA's home health program, or how the pandemic may have affected Veterans’ home health use. Qualitative studies describe barriers to providing home‐based services to rural Veterans, including long travel distances and limited service availability in some areas. ¹⁴ , ¹⁵ , ¹⁶ No studies have quantitatively examined rural‐urban differences in use of home health among Veterans. To address this gap, we examined whether home health utilization differs among rural and urban Veterans. Based on prior findings, we hypothesized that, compared to urban Veterans, rural Veterans would be less likely to use any home health services and that they would receive fewer visits, particularly during the pandemic, which is known to have affected home health use. ¹⁷

METHODS

This study was approved by the University of Iowa Institutional Review Board and followed the STROBE reporting guidelines for cohort studies. ¹⁸

Sample and data sources

This was a retrospective cohort study involving national VA administrative data for 2019–2021 (January 1, 2019 to December 31, 2021). The cohort included Veterans who were ages ≥65 at baseline, enrolled in VA healthcare, with ≥2 outpatient visits or ≥1 inpatient visit in 2018.

We identified Veterans’ demographic and clinical characteristics and service utilization using data from the VA Corporate Data Warehouse. Rurality was determined using geocoded residential addresses available through the VA Planning Systems Support Group dataset. Frailty scores were obtained from the Geriatrics and Extended Care Data & Analysis Center. ¹⁹

Variables

Dependent variables

Home health utilization outcomes were (1) receipt of any home health during the study period (yes/no) and (2) among Veterans who had received ≥1 visit, the number of visits received.

Independent variables

Our main independent variable was residential rurality. We used the definition of rurality used by the VA Office of Rural Health (urban, rural, or highly rural) based on Rural‐Urban Commuting Area Codes (RUCA). ²⁰ We excluded Veterans residing on the US insular islands (territories) because these regions have unique challenges related to home health access and because these areas are not coded using RUCA, thus preventing us from assigning rurality categories for Veterans living in these areas. ²¹ Due to the small number of highly rural Veterans, we grouped rural and highly rural Veterans together.

Additional variables included those previously identified as determinants of home health use ²² , ²³ , ²⁴ : age, sex, race, ethnicity, marital status, frailty, comorbidities, several specific diagnoses (dementia, spinal cord or traumatic brain injury, substance use disorder, and depression), and VA enrollment priority. We also controlled for VA Community Care Network region, the structure through which the VA purchases home health services from community agencies.

Given the small number of Veterans who identified as Asian, American Indian/Alaska Native, other race, or more than one race, we combined these groups into a single “other race” category for analysis. We assessed frailty using the JEN Frailty Index, grouping Veterans into three frailty categories based on categories previously validated within the VA patient population. ²⁵ We used the Gagne Comorbidity Score to quantify Veterans’ mortality risk, categorizing Veterans into four groups based on the overall sample distribution (Appendix A displays characteristics of the full sample). ²⁶ VA enrollment priority levels are based on a Veteran's severity of service‐connected disability and income. ²⁷ Consistent with prior research using VA priority levels as a proxy for economic and disability status, we grouped Veterans into four categories. ²⁸

Analysis

We first conducted descriptive analyses, comparing rural and urban Veterans on demographic and clinical characteristics, and home health utilization. After assessing variables for collinearity, we used a negative binomial hurdle model to estimate Veterans’ use of home health, adjusting for demographic and clinical characteristics. The negative binomial hurdle model is a two‐part model for handling count data with excess zeros and overdispersion. ²⁹ , ³⁰ This modeling strategy enabled us to estimate the probability of a Veteran receiving any home health and number of visits received. We controlled for all demographic and clinical characteristics described previously. Sensitivity analyses examined whether utilization trends were consistent for rural and highly rural Veterans (vs. urban) and across study years (Appendix B). We dropped cases with any missing data (≈5%) from the modeling analyses. Analyses were conducted using SAS Software, Version 9.4. ³¹

RESULTS

Unadjusted home health utilization and Veteran characteristics

The overall cohort included 2.9 million Veterans (Appendix A). Overall, a small percentage of Veterans used home health (3.7%; n = 107,229), a third of whom (33.1%) lived in rural areas (Table 1). In unadjusted analyses, rural Veterans were slightly less likely than urban Veterans to use any home health (3.4% vs. 3.9%; p < 0.001). Among Veterans who received ≥1 visit, there was no statistically significant difference for rural versus urban Veterans in median number of visits over the 3 years (6 visits for both groups; p = 0.42). The median number of visits was similar for rural and urban Veterans across study years (2019 median: 8 visits for both groups; 2020 median: 3 visits; 2021 median: 3 visits).

TABLE 1.

Baseline characteristics of Veterans ages ≥65 using skilled home health services in 2019–2021, stratified by rurality.

Characteristic	Overall N = 107,229 100%	Rural N = 38,525 33.1%	Urban N = 68,704 64.1%	p Value
Age, mean (SD)	77.1 (8.5)	77.0 (8.3)	77.2 (8.6)	0.01
Sex
Male	96.9%	97.5%	96.6%	<0.001
Female	3.1%	2.5%	3.4%
Marital status				<0.001
Married	54.6%	58.0%	52.8%
Divorced/separated	24.9%	22.6%	26.3%
Single/never married	6.2%	4.6%	7.2%
Widowed	13.7%	14.5%	13.3%
Race				<0.001
White	76.8%	86.0%	71.6%
Black/African American	15.6%	6.4%	20.7%
Other race	2.3%	2.3%	2.4%
Hispanic ethnicity	4.7%	2.3%	6.0%	<0.001
Frailty (JEN‐FI)				<0.001
Low frailty (scores: 0–3)	27.6%	28.0%	27.3%
Moderate frailty (4–5)	31.0%	31.7%	30.6%
High/very high frailty (≥6)	39.9%	38.9%	40.4%
Gagne comorbidity score				<0.001
0 or less	24.0%	25.5%	23.2%
1	16.4%	17.2%	16.0%
2	14.5%	14.8%	14.3%
≥3	45.1%	42.5%	46.5%
Dementia diagnosis	16.1%	13.5%	17.6%	<0.001
Spinal cord or traumatic brain injury diagnosis	1.8%	1.5%	2.0%	<0.001
Substance use disorder diagnosis	15.0%	12.3%	16.5%	<0.001
Depression diagnosis	24.9%	23.8%	25.5%	<0.001
VA enrollment priority				<0.001
High disability (Levels 1, 4)	24.5%	23.6%	25.0%
Low‐moderate disability (2, 3, 6)	23.6%	23.7%	23.5%
Low income (5)	30.8%	31.6%	30.4%
No disability/missing (7, 8)	21.1%	21.1%	21.1%
Home health utilization (unadjusted) – based on overall cohort (N = 2,892,522)
Use of any home health, %	3.7%	3.4%	3.9%	<0.001
Number of home health visits among Veterans with ≥1 visit, median (IQR)	6 (2–15)	6 (2–14)	6 (2–15)	0.42

Note: Missing data include marital status (0.6% missing), race (5.3% missing), ethnicity (3.2% missing), marital status and frailty (1.5% missing). Other race (2.3%) includes Veterans who identify as Asian, American Indian and Alaska Native, or more than one race. Frailty was measured using the multisource JEN Frailty Index (JEN‐FI), which uses claims data to identify Veterans with physical function impairment and risk for institutionalization. ²⁵ Comorbidities were measured using the Gagne Comorbidity Score, which quantifies Veterans’ comorbidity burden and risk of mortality based on clinical conditions. Scores range from −2 to 20. ²⁶ VA enrollment priority levels are based on a Veteran's severity of service‐connected disabilities and income. ²⁷ Similar to other research using VA priority levels as a proxy for economic and disability status, we categorized Veterans as high disability (levels 1 and 4), low‐moderate disability (levels 2, 3, and 6), no disability (levels 7 and 8), or low income (level 5). ²⁸

[Correction added on August 10, 2024, after first online publication: The N value for ‘Home health utilization (unadjusted) ’ in column 1 was added.].

Although statistically different due to the large sample size, rural and urban home health users were not meaningfully different in age (mean 77.0 years for rural vs. 77.2 for urban), sex (97.5% male for rural vs. 96.6% for urban), or VA enrollment priority groupings (e.g., 23.6% high disability for rural vs. 25.0% for urban). Rural Veterans were more likely than urban Veterans to be married (58.0% vs. 52.8%) and less likely to be Black (6.4% vs. 20.7%) or Hispanic (2.3% vs. 6.0%). Rural Veterans were less likely than urban Veterans to be diagnosed with depression (23.8% vs. 25.5%), dementia (13.5% vs. 17.6%), or substance use disorder (12.3% vs. 16.5%). Rural Veterans were also slightly less likely to be highly frail (38.9% vs. 40.4%), and they tended to have slightly lower comorbidity scores.

Rurality differences in home health utilization

After adjusting for Veterans’ characteristics, rural Veterans had 10% higher odds of receiving any home health compared to urban Veterans (model part 1: odds ratio [OR] 1.10; 95% confidence interval [CI]: 1.07, 1.13) (Table 2). Among Veterans who received ≥1 visit (model part 2), rurality was associated with 30% fewer expected visits (incident rate ratio [IRR]: 0.70; 95% CI: 0.68, 0.72). Factors associated with greater likelihood of home health receipt (overall and among rural Veterans when assessed separately) included female sex, being unmarried (single/divorced/widowed) versus married, Black race (vs. White race), Hispanic ethnicity, frailty, and comorbidities.

TABLE 2.

Negative binomial distributed hurdle model. Dependent variable: Receipt of skilled home health visits, 2019–2021.

	Model part 1 (receipt of ≥ visit)		Model part 2 (number of visits)
Characteristic	Adjusted odds ratio (OR)	Confidence interval for OR (95%)	Incident rate ratio (IRR)	Confidence interval for IRR (95%)
Rural (reference: urban)	1.10 ***	1.07, 1.13	0.70 ***	0.68, 0.72
Age	1.07 ***	1.06, 1.07	0.98 ***	0.98, 0.98
Female sex (reference: male)	1.58 ***	1.48, 1.69	0.83 ***	0.77, 0.89
Marital status (reference: married)
Divorced/separated	1.55 ***	1.50, 1.59	1.09 ***	1.05, 1.13
Single/never married	1.29 ***	1.23, 1.35	1.31 ***	1.24, 1.38
Widowed	1.71 ***	1.65, 1.78	1.08 ***	1.03, 1.12
Race (reference: White race)
Black race	1.34 ***	1.30, 1.38	1.22 ***	1.17, 1.26
Other race	1.08 *	1.01, 1.16	0.95	0.87, 1.03
Hispanic ethnicity	1.41 ***	1.33, 1.49	1.26 ***	1.18, 1.34
Frailty (reference: low frailty [JEN‐FI scores ≤ 3])
Moderate frailty (JEN‐FI scores 4–5)	1.62 ***	1.57, 1.66	1.06 ***	1.03, 1.10
High frailty (JEN‐FI scores ≥6)	2.47 ***	2.39, 2.56	1.18 ***	1.13, 1.22
Comorbidities (reference: Gagne score ≤0)
Gagne score = 1	1.50 ***	1.46, 1.54	1.14 ***	1.10, 1.19
Gagne = 2	2.04 ***	1.63, 2.11	1.17 ***	1.12, 1.22
Gagne score ≥ 3	4.01 ***	3.86, 4.16	1.50 ***	1.44, 1.56
Dementia diagnosis	5.70 ***	5.07, 6.41	0.82 ***	0.80, 0.85
Spinal cord or traumatic brain injury	2.09 ***	1.83, 2.38	1.17 ***	1.07, 1.29
Substance use disorder diagnosis	1.63 ***	1.56, 1.69	1.11 ***	1.06, 1.15
Depression diagnosis	1.39 ***	1.34, 1.43	1.07 ***	1.04, 1.11

Note: Other race = Asian, American Indian and Alaska Native, and more than 1 race. These races were combined due to the small number of home health users in these groups. JEN‐FI = JEN Frailty Index, a measure based on claims data to identify Veterans with physical function impairment and risk for institutionalization. ²⁵ Gagne = Gagne Comorbidity Score, which quantifies Veterans’ comorbidity burden and risk of mortality based on clinical conditions. Scores range from −2 to 20. ²⁶ Other variables controlled for in the modeling analyses included VA enrollment priority and VA Community Care Network region.

^* p ≤ 0.05.

^*** p ≤ 0.001.

Sensitivity analyses showed similar results for rural and highly rural Veterans. Compared to urban Veterans, highly rural Veterans were more likely to receive any home health (OR: 1.21; 95% CI: 1.11, 1.32) but they received considerably fewer expected visits (IRR: 0.69; 95% CI: 0.62, 0.77). Although rural versus urban trends in home health use were similar over the 3 years (Appendix B), the rural disparity in number of visits increased during the pandemic. Rurality was associated with 24% fewer expected visits in 2019, 38% fewer visits in 2020, and 36% fewer visits in 2021.

DISCUSSION

After controlling for Veterans’ demographic and clinical characteristics, rural Veterans were slightly (10%) more likely to use any home health services than urban Veterans. Consistent with studies conducted among Medicare beneficiaries, however, risk‐adjusted analyses showed that rural Veterans tended to receive considerably fewer home health visits than their urban counterparts. ¹² , ³² Therefore, the disparity observed appears to be one of depth versus breadth of home health coverage.

Home health use decreased among rural and urban Veterans during the pandemic (with median number of visits dropping from 8 to 3 visits between 2019 and 2020). Although the overall rurality‐related patterns of home health use were consistent over the 3‐year study period, we found that the rural disparity in expected number of visits worsened during the pandemic. Studies outside the VA have demonstrated that the home care workforce is declining relative to the number of adults needing these services ⁵ and that home care workforce shortages are more dramatic in rural versus urban areas since COVID‐19. ³³ Because most of the VA's home health services are purchased from the same community agencies that provide care to Medicare beneficiaries, ¹⁰ the rural disparity in home health visits we observed could be similar among Medicare participants, a question future research could examine. Given the growing need for home health services projected among Veterans over the next decade, ⁷ it will become increasingly important for the VA to identify innovative strategies to meet Veterans’ home health needs, particularly in rural areas. ³⁴ For example, the VA is piloting its own (VA‐delivered) home health program at four midwestern sites with large proportions of rural Veterans. The VA could also partner with the Centers for Medicare and Medicaid Services or other government agencies to expand incentives to support rural home health agencies and the rural home health workforce.

Contrary to evidence that rural Veterans have greater clinical needs than urban Veterans, ¹ , ³⁵ our data showed rural Veterans to be less frail, have lower comorbidity scores, and be less likely to have a dementia diagnosis than urban Veterans. Because we limited our analysis to VA healthcare users ages ≥65 years (vs. all VA enrollees), the clinical needs of our cohort may differ from those of the broader Veteran population. It could also be that rural Veterans are less likely than urban Veterans to have all their clinical needs identified. Limited access to specialty services in rural areas, ³⁶ for example, could contribute to the lower rates of dementia we observed among rural Veterans, a trend similarly documented in Medicare studies. ³⁷ , ³⁸ Further, clinical needs identified outside the VA may not consistently be documented in the VA's electronic health records, ³⁹ which could disproportionately affect rural Veterans because they are more likely than urban Veterans to receive specialty care from community providers. ⁴⁰ Reasons for these findings could be explored in subsequent studies, including whether under‐diagnosis of conditions like dementia could limit Veterans’ access to needed home health or other supportive services.

This study has several limitations. First, although we used a highly detailed dataset from the nation's largest integrated healthcare system, ⁴¹ we focused solely on home health services accessed through the VA. Many VA patients, however, use home health services through Medicare. ⁴² , ⁴³ Further, many frail, older Veterans receive home‐based primary care, ⁴⁴ which may lessen the need for other home health services. Therefore, our results likely underestimate Veterans’ home health use. Second, our focus on current utilization trends meant that our study period overlapped with the COVID‐19 pandemic, which affected home health use. ¹⁷ , ⁴⁵ Subsequent research could examine whether the trends we observed remain consistent beyond 2021, and whether rural disparities in home health visits have remained similar, lessened, or worsened. Third, although we included marital status as a rough proxy for whether a Veteran lived alone, VA administrative data does not currently include a measure of Veterans’ living arrangements or whether they had a family caregiver, so we were unable to examine how these factors may influence Veterans’ use of home health. Finally, because we excluded Veterans living in insular islands, we do not know how rurality may affect home health use among island‐dwelling Veterans, such as those living in Guam or Puerto Rico. Future work examining home health access among these, and other subpopulations of Veterans is needed.

CONCLUSION

This study makes an important contribution to understanding rurality‐based differences in home health use by demonstrating that, although rural Veterans are slightly more likely to receive any home health than urban Veterans, they tend to receive considerably fewer visits. Future research could use qualitative methods to examine reasons for this difference, and whether rural Veterans experience worse outcomes—such as more hospitalizations or greater likelihood of nursing home placement—as a result. As the gap between the need for home health and available services widens, ⁵ , ⁷ ongoing attention is needed to ensure rural‐dwelling older adults—and subgroups of rural dwellers—receive equitable care.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to declare.

Supporting information

Supporting Information

Supporting Information

Davila H, Mayfield B, Mengeling MA, et al. Home health utilization in the Veterans Health Administration: Are there rural and urban differences?. J Rural Health. 2025;41:e12865. 10.1111/jrh.12865