Healing at Home: Receipt of Home Health Care and Patient-centered Outcomes among Older Adults after Acute Myocardial Infarction

Abstract

Objective:

To evaluate the impact of receipt of home health care (HHC) on the patient-centered outcomes of days not at home (DNAH) and risk of decline in physical and mental health status among older adults hospitalized with acute myocardial infarction (AMI).

Design:

Secondary analysis of data from a national prospective cohort study.

Settings and Participants:

Adults age ≥75 years discharged home after an AMI hospitalization and linked to Medicare data.

Methods:

The primary exposure was receipt of HHC by 14 days post-discharge. Days not at home (DNAH) within 180 days after discharge was defined as the count of days spent deceased or in a hospital, skilled nursing facility, or inpatient rehab facility. Health status decline was defined as a decrease ≥5 points on the SF-12 from baseline to 180 days. Inverse probability of treatment-weighted generalized linear models were used to evaluate associations between receipt of HHC and these outcomes.

Results:

491 (26.6%) of 1846 participants received HHC within 14 days; HHC recipients were older and had greater clinical and functional comorbidity. HHC was not associated with DNAH (IRR=1.24, 95% CI 0.99-1.56) nor risk of decline in mental health status (OR=0.83, 95%CI 0.64-1.07), but was associated with substantially reduced risk of decline in physical health status (OR=0.70, 95% CI 0.56-0.89).

Conclusions and Implications:

More than one-quarter of AMI patients age ≥75 discharged home received HHC. Patients who received HHC, who were older and sicker on average, were less likely to experience clinically significant declines in physical health status but had similar rates of mental health status decline and DNAH as those who did not receive HHC. This work suggests that providers consider the health outcome goals of older AMI patients and prescribe HHC to those who prioritize preservation of physical health status.

summary:

Receipt of home health care was associated with reduced risk of clinically significant decline in physical health status among older myocardial infarction patients at 180 days post-discharge.

Background

Despite comprising only 6% of the population,¹ adults age ≥75 comprise up to 40% of all hospitalizations for acute myocardial infarction (AMI).² Due to advances in acute treatment of myocardial infarction, such as percutaneous intervention, short-term survival after AMI has increased dramatically.³ However, challenges remain with longer term management of AMI in older adults, evidenced by high rates of mortality,^4,5 readmission,⁶ institutionalization,⁷ decline in physical function,^8,9 and quality of life^8,10 in the months after discharge. Moreover, the high costs of care for AMI, attributed to both the index admission and frequent unplanned readmissions, places AMI among the most costly conditions covered by Medicare.¹¹ This financial burden has prompted a call by the Centers for Medicare and Medicaid Services to improve post-discharge outcomes among older adults hospitalized with AMI. Home health care (HHC), the most common type of post-acute care used by older AMI patients,^12,13 holds promise as a bridging strategy to facilitate the hospital-to-home transition and improve outcomes in the post-discharge period, while decreasing costs associated with inpatient post-acute care. HHC, which is provided by Medicare-certified HHC agencies and can include skilled nursing, physical and occupational therapy, and home health aide and social services, is delivered at home and aims to support patients and maintain or improve function and monitor symptoms after discharge. HHC is often aligned with patient preferences to go home (vs. to an inpatient facility) after a hospitalization and is typically delivered at lower cost than inpatient post-acute care. Despite the widespread use of HHC after AMI,^12,13 little empirical evidence exists to support its efficacy in improving clinical and patient-centered outcomes. The few existing studies in AMI patients that have evaluated the association of receipt of HHC with risk of readmission after hospitalization for AMI^14,15 included younger populations and focused exclusively on ST-elevation myocardial infarction, which may make their findings less generalizable to older, medically complex older adults who overwhelmingly present with non-ST elevation MI (NSTEMI).² Other studies evaluating HHC have been limited to patients undergoing coronary artery bypass grafting (CABG; with or without concomitant AMI),^16,17 which is a less favored treatment strategy among medically complex older adults⁷ compared with percutaneous coronary intervention (PCI) or medical management. No studies to date have evaluated the impact of HHC in a broad population of older patients hospitalized with AMI on important patient-centered outcomes such as time spent at home or health status,¹⁸ which older adults often value as much as, or more than, clinical outcomes such as staying alive.¹⁹

A primary concern when evaluating the impact of HHC on outcomes is the potential for confounding by indication, i.e., patients who receive HHC are often sicker and more vulnerable than patients discharged home without home health care,^20,21 and therefore more likely to experience poor outcomes. Prior studies have noted limitations in their ability to account for heterogeneity in disease severity, comorbidities, and functional vulnerabilities between patients who do and do not receive HHC.^16,17 This study utilizes data from the ComprehenSIVe Evaluation of Risk Factors in Older Patients with AMI (SILVER-AMI) study (NIH R01HL115295), a unique prospective cohort study that was designed to capture the clinical heterogeneity and functional capabilities critical for controlling for confounding by indication.

The objective of this study was to evaluate the impact of home health care on days spent not at home and change in health status in the 180 days after discharge from AMI hospitalization. Days spent not at home, which represents the number of days spent hospitalized, in the emergency department, in a skilled nursing facility, inpatient rehab, or dead, is an outcome that combines risk of morbidity and mortality and represents older adults’ preference to spend their days alive and at home rather than in health care facilities.²² Changes in health status measures shifts in an older adult’s physical or mental condition and reflects changes to symptoms, functioning, and overall well-being. These important patient-centered outcomes are increasingly recognized²³ as critical determinants of health care quality,²⁴ particularly for older adults.

Methods

Study cohort and data sources

This secondary data analysis used data from the SILVER-AMI study linked to fee-for-service claims and Medicare Advantage encounter data from the Centers for Medicare and Medicaid Services (CMS). The SILVER-AMI study has been previously described.²⁵ Briefly, SILVER-AMI was a prospective cohort study of 3,041 adults age ≥75 years hospitalized with AMI recruited from 94 academic and community hospitals across the United States from 2013 to 2016. Site coordinators reviewed hospital admission records daily to identity potentially eligible participants and performed medical record review to confirm AMI diagnosis in accordance with the Third Universal Definition of Myocardial Infarction.²⁶ Patients were ineligible if they had initial troponin elevation >24 hours after admission, were transferred from an outside hospital after >24 hours, experienced AMI as a result of an in-hospital procedure, were incarcerated, or were unable to provide informed consent with no proxy available. Study protocols were approved by the Institutional Review Boards at Yale and all participating sites. Participants underwent a structured interview and physical assessment during AMI hospitalization to collect information on mobility, activities of daily living, cognitive function, hearing, vision, demographics, and psychosocial factors. Clinical characteristics, including comorbidities, cardiovascular risk factors, and details of the MI hospitalization (e.g., length of stay, treatments, complications) were abstracted from the medical record.

For the current study, SILVER-AMI participants were included in the study if they were discharged alive to home (private residence, independent or assisted living facility) and were successfully linked to Medicare data. We obtained Medicare Part A and Part B claims and Medicare Advantage encounter data for each SILVER-AMI participant from the month of the index AMI admission through 12 months post-discharge (spanning 2013 to 2017). CMS files obtained for these analyses included Medicare Provider Analysis and Review (MedPAR); Outcome and Assessment Information Set (OASIS), a mandatory assessment used to evaluate the quality of care provided by all Medicare-certified HHC agencies; Inpatient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI); inpatient; outpatient, skilled nursing facility, home health agency, minimum data set, and hospice research identifiable files; as well as the Master Beneficiary Summary File. SILVER-AMI participants were matched to their Medicare data by CMS via patient identifiers (e.g., social security number, date of birth, last name, index admission date). We allowed fuzzy matching (e.g., one day off) for birth dates or admission dates. Participants for whom we could not link Medicare data were excluded (n=694). We also excluded participants if they died during the index admission (n=35) or were discharged to a location other than home (i.e., skilled nursing facility, inpatient rehab facility, hospice) after the index admission (n=445). The reporting of study procedures conforms to STROBE guidelines.

Study Measures

Receipt of HHC in the early post-discharge period

Our primary independent variable was receipt of HHC within 14 days of discharge from the index AMI admission. We used OASIS data to confirm that a participant received a HHC evaluation as well as the start and end dates of the HHC episodes. We used data from the home health agency file to determine the number of HHC visits and home health care type (i.e., skilled nursing, physical therapy, occupational therapy, home health aide, social services, and speech therapy). We performed quality checks to confirm that the home health visit dates corresponded to the index hospitalization discharge dates.

Outcomes

We evaluated outcomes at 180 days after discharge from the index AMI hospitalization to allow sufficient time for any impact of home health care (each episode of which lasts for 60 days) on outcomes to manifest, and also because the vast majority of post-AMI adverse events occur during this period.²⁷

Days not at home (DNAH)

We defined days not at home as the number of days within 180 days after the index AMI hospitalization discharge that a participant was deceased, in the emergency department, hospitalized, in a skilled nursing facility, in an inpatient rehab facility, or in institutional hospice. We calculated days dead by subtracting the date of death, found in the Medicare Beneficiary Summary File, from the ending date of each participant’s 180-day follow-up window. We calculated hospital days by summing the days of all inpatient days in MedPAR or the inpatient claims and encounter files within 180 days of the index admission discharge; ED visit days were calculated separately. We calculated skilled nursing facility days by summing the number of days spent in a skilled nursing facility using data from MedPAR or minimum data set files and the number of days spent in inpatient rehab facilities using data from MedPAR or IRF-PAI files. We calculated days spent in inpatient hospice using data from the hospice claims files. We then calculated the sum of all these domains to create the count of DNAH within the 180-day follow-up period. We used DNAH, rather than days at home, as our primary outcome, as this inverted operationalization facilitated use of zero-inflated negative binomial regression for our analyses.

Health Status

Patient-reported health status was assessed using the SF-12.²⁸ The SF-12 was administered in the original SILVER-AMI study during hospitalization (asking participants to report upon pre-AMI health status) and at six months post-discharge, yielding physical and mental component scores ranging from 0 to 100 (higher indicates better health status). We calculated change in physical health status and mental health status scores from hospitalization to six months post-discharge by subtracting each participant’s score at baseline from their score at six months. We defined clinically relevant decline in physical or mental health status according to the minimally important clinical difference of ≥5 points, which corresponds to a half standard deviation change (i.e., moderate effect size)²⁹ and with precedent use in studies of cardiac populations.^30,31

Covariates

Demographic and lifestyle information (age, sex, race, ethnicity, marital status, education, smoking status, high alcohol use (>7 drinks/week for females or >14/week for males or >4 drinks on a single day in the past month) were collected in the original SILVER-AMI study via self-report or medical record review. Data on prior medical history, comorbidities, and AMI hospitalization characteristics, including AMI treatment (medical management only, cardiac angiography only, percutaneous coronary intervention, coronary artery bypass surgery), were extracted from participants’ medical records. Mobility impairment,³² cognitive status,³³ essential activities of daily living,^34,35 grip strength,³⁶ vision,³⁷ and hearing³⁸ were evaluated with performance-based assessments or self-report. Psychosocial characteristics of perceived health,²⁸ depression,³⁹ and social support⁴⁰ were captured via validated questionnaires. For home health care recipients, characteristics of the home health care episode, including number of visits and type, were extracted from OASIS and home health care claims files. Health insurance status (i.e., traditional vs. Medicare Advantage) at the time of discharge from the index admission was ascertained from the Medicare Beneficiary Summary File. Lastly, we used the SILVER-AMI mortality risk score,⁴¹ previously developed using SILVER-AMI data to estimate mortality risk within 180 days of discharge after AMI, to account for heterogeneity in survival probability.

Statistical Analysis

Descriptive statistics with standardized mean differences⁴² were used to examine differences in demographic, clinical, and functional characteristics between participants who did and did not receive HHC within 14 days of hospital discharge. We considered standardized mean differences (SMD) ≥0.2 to indicate non-negligible differences between groups, which corresponds to a small effect size.⁴³ Characteristics of the home health episode were described for HHC recipients only.

To assess the association between receipt of HCC and DNAH, we estimated a series of count models. First, we estimated Poisson, zero-inflated Poisson, negative binomial, zero-inflated negative binomial, and generalized linear with binomial link models, using Akaike’s and Bayesian Information criteria (AIC and BIC) from each model to identify which model best fit our data. The model with lowest values of AIC and BIC, the zero-inflated negative binomial model, was used for analysis. To assess whether HHC was associated with a decline of ≥5 points in the mental or physical SF-12 scores, we estimated analogous logistic regression models.

To address residual confounding by indication for receipt of home health care, a propensity model was estimated using the LASSO procedure for model selection to generate a probability of HHC receipt.^44,45 The propensity model included all variables in Table 1, unless otherwise noted. Inverse probability of treatment weighting (IPTW)⁴⁶ was applied in all models along with covariates that remained unbalanced (standardized mean difference >0.2) after application of IPTWs. All analyses were performed in SAS 9.4 and Stata 16 (StataCorp, College Station TX). We used a critical value of alpha = 0.05 to determine significance.

Table 1.

Participant characteristics by receipt of home health within 14 days of discharge

Characteristic	Participants who did not receive HHC N=1373	Participants who received HHC N=491	SMD	SMD after IPTW
Demographics
Age, mean (SD)	80.8 (4.6)	81.9 (5.0)	0.222	0.124
Sex, male, n(%)	849 (61.8)	229 (46.6)	0.309	0.090
Race, n(%)			0.236	0.136
White	1215 (88.5)	434 (88.4)
Black	84 (6.1)	46 (9.4)
Other	74 (5.4)	11 (2.2)
Hispanic ethnicity, n(%)	35 (2.5)	20 (4.1)	0.085	0.043
Education, ≤high school, n(%)	720 (52.4)	313 (63.7)	0.235	0.094
Marital status, married, n(%)	793 (57.8)	214 (43.6)	0.279	0.102
Lives alone, n(%)	462 (33.6)	209 (42.6)	0.184	0.149
Insurance type at discharge, n(%)			0.076	0.092
Traditional Medicare	864 (62.9)	328 (66.8)
Medicare Advantage	504 (36.7)	163 (33.2)
Annual income <$50K	771 (56.2)	310 (63.1)	0.208	0.123
Home health care agency density (per 10K population >age 65)			0.283	0.207
≤10	102 (7.4)	11 (2.2)
11-30	272 (19.8)	86 (17.5)
31-60	347 (25.3)	127 (25.9)
61-120	318 (23.2)	134 (27.3)
121-270	203 (14.8)	95 (19.3)
≥270	99 (7.2)	34 (6.9)
Clinical Characteristics
Myocardial Infarction type, n(%)
STEMI	394 (28.7)	126 (25.7)	0.068	0.052
NSTEMI	979 (71.3)	365 (74.3)
Length of stay, median[IQR]	3 [2-5]	6 [4-9]	0.806	0.104
Revascularization
None	165 (12.2)	71 (14.5)	0.519	0.070
Cardiac catheterization only	243 (17.7)	80 (16.3)
Percutaneous coronary intervention	916 (66.7)	249 (50.7)
Coronary artery bypass graft	49 (3.6)	91 (18.5)
In-hospital complications
Acute kidney injury	197 (14.3)	142 (28.9)	0.359	0.030
New heart failure	123 (9.0)	84 (17.1)	0.244	0.066
New arrhythmia	201 (14.6)	106 (21.6)	0.181	0.101
White blood cell count, median [IQR]	9.6 [7.8-12.2]	11.1 [8.7-14.7]	0.264	0.061
Medical History
Charlson score, median [IQR]	3.8 (2.7)	3.3 (2.6)	0.177	0.102
Heart failure at admission	206 (15.0)	115 (23.4)	0.215	0.117
Chronic kidney disease at admission	793 (57.8)	304 (61.9)	0.083	0.035
Diabetes at admission	479 (34.9)	185 (37.7)	0.058	0.019
Chronic obstructive pulmonary disease at admission	168 (12.2)	78 (15.9)	0.105	0.194
Impairment in ≥1 essential activities of daily living	117 (8.5)	81 (16.5)	0.243	0.063
Mobility impairment, Timed Up and Go*
<15 seconds (normal)	560 (40.8)	87 (17.7)
≥15 to <25 seconds (mild impairment)	334 (24.3)	89 (18.1)	0.702	0.298
≥25 seconds (moderate impairment)	167 (12.2)	117 (23.8)
Unable to complete (severe impairment)	84 (6.1)	76 (15.5)
Unable to walk ¼ mile before admission, n(%)	389 (28.3)	207 (42.2)	0.296	0.068
>2 falls in past year, n(%)	216 (15.7)	105 (21.4)	0.151	0.057
Grip strength weakness,† n(%)	752 (54.8)	308 (62.7)	0.210	0.044
Hearing impairment	742 (54.0)	252 (51.3)	0.052	0.014
Visual impairment	451 (32.8)	201 (40.9)	0.170	0.036
Cognitive impairment	152 (11.1)	100 (20.4)	0.272	0.080
Ever smoker, n(%)	784 (57.1)	277 (56.4)	0.014	0.013
High alcohol use,‡ n(%)	82 (6.0)	17 (3.5)	0.119	0.094
Depression (PHQ), median [IQR]	3 [1-6]	4 [2-8]	0.276	0.051
Social support (MOS-SSS), median [IQR]	24 [20-25]	23 [19-25]	0.108	0.098
Physical health status (SF-12), median [IQR]	43 [36 -50]	40 [33-49]	0.234	0.068
Mental health status (SF-12), median [IQR]	55 [48-60]	54 [45-60]	0.165	0.051
6-month SILVER-AMI mortality risk score,§ median [IQR]	3.5 [1.7-7.3]	6.6 [2.7-12.5]	0.401	0.103

Descriptive characteristics reported as n(%), mean(SD), or median(IQR).

SMD= standardized mean difference. IPTW=inverse probability of treatment weighting. STEMI= ST-elevation myocardial infarction. NSTEMI= non-ST elevation myocardial infarction. PHQ= Patient Health Questionnaire. MOS-SSS= Medical Outcomes Study Social Support Scale. SF-12= Short Form 12.

^*Mobility impairment categories adopted from prior work in SILVER-AMI⁶²

^†Grip strength weakness defined as <28.5 kg for men and <18.5 for women.

^‡High alcohol use defined as >14 drinks per week for men and >7 drinks per week for women or any occasion of >4 drinks on a single day in the past month.

^§Details of SILVER-AMI mortality risk score can be found at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7695040/.

Results

We successfully matched 2312 of the 3,006 SILVER-AMI participants who were discharged alive from the index AMI hospitalization to Medicare data. After excluding 374 participants who were not discharged home and three participants who died within 48 hours of discharge (i.e., were not in eligible window to receive HHC evaluation) our final analytic cohort comprised 1864 older adults. Of those, 491 (26%) received HHC within 14 days of discharge from their index AMI admission. The vast majority of home health recipients for which we could determine HHC type (n=376 of 491) received skilled nursing care (96.8%), followed by physical therapy (66%), occupational therapy (22.3%), and home health aide visits (12.5%; Figure 1).

Figure 1.

Type of Home Health Care Services Received among Home Health Care Recipients

Differences in demographic, clinical, and psychosocial characteristics between patients who did and did not receive HHC within 14 days of AMI hospitalization discharge are shown in Table 1. HHC recipients were older (mean age=81.9 (5.0) vs. 80.8 (4.6) years, SMD=0.22), more likely to be female (53.4% vs 38.2%, SMD=0.31), more likely to be Black or African American (9.4% vs. 6.1%, SMD=0.24), less likely to have education beyond high school (35.4% vs. 46.9%, SMD=0.24), less likely to be married (43.6% vs 57.8%, SMD=0.28), and more likely to live alone (42.6% vs. 33.6%, SMD=0.18).

For clinical characteristics, AMI type did not differ substantially between those who did and did not receive HHC (STEMI: 25.7% vs. 28.7%, SMD=0.07). HHC recipients were less likely to have undergone PCI (50.7% vs. 66.7%) and more likely to have undergone CABG (18.5% vs. 3.6%, SMD=0.52) during their AMI hospitalization than patients who did not receive HHC. HHC recipients had longer lengths of stay on average (median [IQR]= 6 [4-9] vs. 3 [2-5] days, SMD=0.81), potentially attributable to receipt of CABG or experiencing complications such as acute kidney injury (28.9% vs. 14.3%, SMD= 0.24), heart failure (17.1% vs 9.0%, SMD=0.24), or arrhythmia (21.6% vs. 14.6%, SMD=0.18). Rates of some, but not all, pre-existing comorbidities were higher among HHC recipients than those who did not receive HHC, evidenced by Charlson comorbidity scores of 3.8 (SD=2.7) and 3.3 (SD=2.6, SMD=0.18) and history of heart failure (23.4% vs. 15.0%, SMD=0.22). Notably, 180-day all-cause mortality risk, measured with the SILVER-AMI risk score, was substantially higher among HHC recipients (median [IQR]= 6.6% [2.7-12.5%] than those who did not receive HHC (3.5% [1.7-7.3%], SMD=0.40).

Differences in prevalence of geriatric vulnerabilities between those who did and did not receive HHC were substantial. HHC recipients demonstrated higher rates of mobility impairment (39.3% vs. 18.3%, SMD=0.70), cognitive impairment (20.4% vs. 11.1%, SMD=0.27), grip strength weakness (62.7% vs. 54.8%, SMD=0.21), and impairment in essential activities of daily living (16.5% vs. 8.5%, SMD=0.24). HHC recipients also reported lower mean physical health status (40.5 (10.0) vs. 42.8 (9.8), SMD=0.23) and mental health status (51.7 (10.5) vs. 53.3 (9.5); SMD=0.17) than those who did not receive HHC at baseline, although these differences were smaller than the minimally clinically important difference. All variables, with the exception of mobility impairment, were adequately balanced after IPTW.

More than half (n=978, 52.3%) of participants spent at least one day not at home in the six months after discharge. HHC recipients were more likely to spend at least one day not at home (n=302, 61.5%) than those who did not receive HHC (n=676, 49.1%, p<0.001), and the median number of DNAHs among participants with at least one DNAH was greater among HHC recipients (8 [IQR 3-37] vs 5 [IQR 2-19], p<.001). Proportionally, HHC recipients spent on average 11.7% of the 180 days after discharge not at home, compared with 7.0% of the 180 days after discharge spent not at home among participants who did not receive HHC (p<0.001; Supplemental Table 1). Proportions of days deceased (5.3% vs. 3.1%, p<0.001), days in short-stay hospitals (3.0% vs 2.0%, p<0.001), days in ED (0.3 vs. 0.2, p<0.001), days in SNF (2.1% vs. 1.1%, p<0.001), and days in home hospice (1.5% vs. 0.5%, p<0.001) were significantly higher among HHC recipients relative to participants who did not receive HHC, but proportions of days spent in inpatient rehab, long-term acute care hospital, or inpatient hospice were not significantly different. After applying IPTWs and adjusting for the unbalanced covariate (mobility impairment), utilization of HHC was not associated with DNAH within 180 days of discharge (IRR=1.24, 95% CI 0.99-1.56; Figure 2). The average number of days not at home estimated by the IPTW, multivariable-adjusted model was 2.8 (95% CI 2.6-2.9) days for HHC recipients and 2.5 (95% CI 2.4-2.7) days for those who did not receive HHC, results that were neither statistically (p=0.06) nor clinically significant in accordance with a minimally clinically important difference threshold of approximately 9 days for a 180-day period.⁴⁷

Figure 2.

Association of Receipt of Home Health Care with Days Not at Home and Decline in Health Status 180 Days after AMI Hospitalization

In unadjusted analyses, clinically relevant decline (i.e., ≥5 points) in physical health status occurred in 32.1% of HHC recipients and 33.7%% of those who did not receive HHC (p=0.56). After applying inverse probability of treatment weights and controlling for the residual unbalanced covariate, patients who received HHC had 30% lower odds of reporting a decline in physical health status (OR=0.70, 95%CI 0.56-0.89; Figure 2). Clinically significant decline in mental health status occurred in 39.3% of HHC recipients and 34.6% of participants who did not receive HHC (p=0.09). After applying inverse probability of treatment weights and controlling for residual unbalanced covariates, patients who received HHC had 17% lower odds of reporting decline in mental health status, though this was not statistically significant (OR=0.83, 95% CI 0.64-1.08, NS; Figure 2).

Discussion

In this prospective cohort study linked to Medicare data, we found that more than one-quarter of older adults hospitalized with AMI who were discharged to the community received HHC within two weeks of discharge. In adjusted analyses incorporating inverse probability of treatment weights, we found no difference between those who received HHC and those who did not for the outcomes of DNAH or decline in mental health status. Receipt of HHC was associated with a significant decrease in the risk of decline in physical health status.

To our knowledge, this is the first study to evaluate the association of HHC with DNAH after AMI. Previous work focused on the use of HHC to prevent short-term (i.e., 30-day) unplanned health care utilization in select subpopulations (STEMI patients or CABG recipients), finding reduced readmissions and emergency department visits^14,16 but slightly higher mortality rates.¹⁶ In contrast, when looking over a longer period (i.e., 180 days), we did not find that HHC receipt was associated with statistically or clinically significant differences in the count of DNAH, which comprise days lost due to death or institutionalization. Comparisons to the prior literature must be made with caution, due to numerous variations in methods, including variations in outcomes, follow-up times, selection criteria, confounding variables that were accounted for, and importantly, the more advanced age of our cohort relative to those of previous studies.

There are several possible interpretations of our finding of no benefit to DNAH among recipients of home health care. It is possible that HHC was not effective in reducing DNAH in our relatively aged and multimorbid cohort of AMI survivors, which corroborates with some prior literature showing no benefits of HHC in terms of risk of readmission,⁴⁸ institutionalization,⁴⁹ or mortality risk.¹⁶ A null finding may also represent an amalgam of two conflicting phenomena- (1) the prevention of DNAH that are attributable to ambulatory-care sensitive conditions and (2) the addition of DNAH due to identification of medical needs necessarily requiring attention in the ED or hospital. Furthermore, it has been suggested the medico-legal concerns may prompt HHC providers to recommend return to hospital for patients with clinical concerns, due to limited resources and training to perform full diagnostic workups and a desire to avoid adverse outcomes occurring at home.⁴⁸ Lastly, it is also possible, although less likely, that our finding is driven by factors associated with increased risk of mortality and health care utilization that are not contained in our comprehensive dataset; while use of data from SILVER-AMI study allowed us to account for many important differences in medical, functional, and social needs among HHC recipients (i.e., lower education, less likely to be married, more likely to live alone, more functional and cognitive impairments, higher risk of six-month mortality), it is possible that additional, unmeasured, and influential characteristics, such as detailed information on informal (family) care support or adherence to secondary preventive activities, could be confounding the association between receipt of HHC and DNAH. It is not possible to test this theory with the available data, as propensity-based methods can only address imbalance in observed characteristics. A more robust study design (e.g., randomization) would be needed.

Our finding that HHC was associated with lower risk of physical health decline (comparing physical health at six months post-discharge to that prior to AMI hospitalization) among our older AMI patients is encouraging and in line with studies of general populations (i.e., not AMI-specific, not recently hospitalized) in the United States⁵⁰ and Korea⁵¹ that reported similar benefits of HHC among older adults, including improvements in mobility and self-care⁵⁰ and improvements in activities of daily living, general health, and quality of life.⁵¹ This finding is especially important in light of previous work showing that one-third of older adults experiencing an AMI hospitalization experience a decline in physical health status.³¹ HHC may prevent physical health status decline by providing physical therapy, assistive devices, and home modifications to restore or maintain physical health and improve physical abilities that have been decremented as a direct result of the AMI^52,53 or the iatrogenic impacts of hospitalization on physical health.⁵⁴ Physical or occupational therapy may remediate losses in strength, mobility, balance, and ability to perform self-care tasks, and skilled nursing may promote recovery through managing medications and performing wound care, while home health aides may promote recovery through ensuring proper nutrition, grooming, home upkeep, and activity levels. Given that a primary goal of HHC is to restore or maintain patients’ physical and mental functioning, independence, and quality of life,⁵⁵ all key components of the holistic construct of health status, our results add evidence in support of HHC improving this important patient-centered outcome, at least in terms of physical health status, for older AMI patients.

Our mixed findings for the benefits of HHC after AMI hospitalization come at a time of increased awareness and consideration of patient values and goals in supporting shared health care decision-making.⁵⁶ Initiatives such as the Patient Priorities Care approach encourage clinicians to discuss with patients what “matters most” to them for their post-hospitalization recovery⁵⁷ and develop a care plan that supports achieving the patient’s personal health goals. With the new evidence provided in this study, clinicians have additional information to share with patients regarding expectations for staying alive and out of the hospital, as well as for maintaining physical and cognitive health, associated with receipt of HHC. While HHC may not deliver substantial benefits with regard to staying alive and out of the hospital compared to usual care, it may represent an important tool for patients who prioritize maintaining their physical health status, and potentially preventing downstream institutionalization.⁵⁸

Our study has a number of important strengths. First, our use of the unique SILVER-AMI data source, enriched with older AMI patients that contained comprehensive information on aging-related characteristics, enabled us to account for important sources of residual confounding by indication (e.g., functional and cognitive status) reported by previous investigations.¹⁶ Second, we explored the innovative outcomes of DNAH and change in health status within 180 days after AMI, which correspond to older adults’ medical treatment priorities and are hailed as better performance metrics than mortality or readmissions⁵⁹ but are nonetheless understudied in cardiovascular outcomes research.⁶⁰ Third, we employed a rigorous analytic approach to minimize observed confounding by indication.

This study has several limitations. First, our propensity score was not able to fully balance one potentially influential covariate (i.e., mobility impairment), and there are likely other unobserved characteristics not captured in our data set (e.g., provider referral preferences, access to outpatient care, and other psychosocial factors) that impact both the probability of receiving HHC and the risk of experiencing our outcomes. We did adjust for the covariate that was not fully balanced by the propensity score. Second, we were not able to empirically evaluate associations of HHC with our outcomes according to the dose or type of HHC received. Third, we were not able to discern the appropriateness of the health care utilization included in our DNAH outcome. Lastly, the frequency of DNAH in both study groups was quite low, i.e., only two to three days not at home over 180 days of follow-up, on average. This frequency was lower than expected and lower than the frequency of days not at home in other cohorts,⁶¹ suggesting that the included patients may have been healthier than typical cohorts of older patients post-AMI. This low rate may also have made it more difficult to detect differences.

Conclusions and Implications

In conclusion, we found that more than one-quarter of adults age ≥75 hospitalized of AMI and discharged home received HHC within two weeks of discharge, and that those who received HHC were less likely to experience clinically significant declines in physical health status in the six months after discharge, relative to pre-admission health status. However, HHC recipients had similar rates of mental health status decline and DNAH to those who did not receive HHC. This work suggests that providers consider the health outcome goals of their older patients hospitalized with AMI and prescribe HHC to those who prioritize preservation of physical health status.