Caregiver Burden at the Onset of Acute Hospital‐At‐Home

ABSTRACT

Background

Caregiver burden at the onset of acute Hospital‐at‐Home (HaH) episodes, particularly in programs where caregiver availability is required for admission, remains insufficiently characterized in real‐world settings.

Participants and Setting

Caregivers of patients admitted to an acute HaH program within Clalit Health Services, Northern District, Israel, between 2023 and 2024.

Methods

We conducted a cross‐sectional observational study of 125 caregivers assessed within the first 48 h of the HaH episode. Caregiver burden was measured using the Caregiver Strain Index (CSI; range 0–13), with high burden defined as CSI ≥ 7. Structured telephone interviews collected caregiver demographics, caregiving context, health‐related quality of life (EQ‐5D‐5L; UK value set), resilience (CD‐RISC‐2), and perceived social support. Associations with high burden were examined using bivariate analyses and hierarchical logistic regression.

Results

High caregiver burden was present in 77 caregivers (61.6%) within the first 48 h of the HaH episode. Prior caregiving experience was more prevalent among caregivers with high burden than among those with low burden (97.4% vs. 50.0%, p < 0.001). High burden was also associated with anxiety/depression on the EQ‐5D‐5L (63.6% vs. 29.2%, p < 0.001), lower resilience (52.0% vs. 79.2%, p = 0.002), and lower perceived social support (59.7% vs. 83.3%, p = 0.006). In multivariable analysis, perceived social support (OR 0.22, 95% CI 0.06–0.74) and high resilience (OR 0.33, 95% CI 0.11–0.95) were associated with lower odds of high burden, whereas paid caregiver assistance was associated with higher odds of high burden (OR 5.23, 95% CI 1.17–23.37).

Conclusions

High caregiver burden is common at the onset of acute HaH care. These exploratory findings suggest that caregiver vulnerability at admission is related to pre‐existing caregiving context and psychosocial resources, supporting systematic caregiver assessment early in the HaH episode to identify individuals who may benefit from targeted support.

Summary

• Key Points

  • ○
    Prevalence: Nearly two‐thirds of caregivers (61.6%) reported high burden (CSI ≥ 7) within the first 48 h of an acute Hospital‐at‐Home episode.
  • ○
    Psychosocial correlates: High burden was associated with emotional distress, poorer health‐related quality of life, lower resilience, and limited perceived social support; in multivariable analysis, perceived social support and high resilience were each independently associated with lower odds of high burden.
  • ○
    Caregiving context: Prior caregiving for the same patient was nearly universal among caregivers with high burden (97.4%), while caregiving‐naïve individuals more commonly reported low or no burden at HaH onset—suggesting that burden at HaH onset largely reflects a caregiving context predating the acute episode.
• Why Does This Paper Matter?

  • ○
    Caregiver burden is clinically visible at the onset of acute Hospital‐at‐Home and should not be inferred from patient diagnoses or program outcomes alone. Systematic caregiver assessment at HaH admission may identify vulnerable caregivers and enable early targeted support.

Among 125 caregivers assessed within 48 h of acute Hospital‐at‐Home (HaH) admission, 61.6% had high burden—nearly all with prior caregiving experience. Perceived social support and resilience were independently associated with lower odds of high burden, identifying psychosocial resources as key targets for caregiver screening at HaH onset.

1. Introduction

Hospital‐at‐Home (HaH) programs are increasingly implemented as effective and safe alternatives to inpatient care for older adults, in response to population aging, multimorbidity, and hospital capacity constraints [1, 2, 3, 4, 5]. Admission‐avoidance HaH refers specifically to active, time‐limited treatment delivered by healthcare professionals in the patient's home for a condition that would otherwise require acute inpatient hospitalization [1]. Acute HaH is therefore distinct from chronic home‐based primary care, post‐acute early discharge schemes, and hospice care, as it involves a sudden transition to high‐intensity medical care in the home setting. This abrupt transition includes daily medical and nursing visits and, in most cases, intravenous therapy administered in the home.

In Israel, acute home hospitalization has been implemented since the early 2000s and expanded substantially during the COVID‐19 pandemic, reaching over 30,000 national admissions in 2023 [6]. This transition creates a structural dependency on caregivers—defined in this study as the primary informal family member or relative assuming responsibility for the patient during the HaH episode (hereafter: caregiver). Caregiver cooperation is a prerequisite for admission. Caregivers are therefore not peripheral participants but main actors in care delivery, assuming immediate responsibilities for supervision and day‐to‐day support alongside the clinical management provided by the HaH team [7]. In Israel, acute HaH is fully covered under mandatory insurance, while long‐term care support remains means‐tested and fragmented—a context in which caregivers routinely carry responsibilities unsupported by the formal system. As HaH programs expand, caregiver capacity becomes a key determinant of program feasibility and safety [3, 8, 9].

The experience of caregivers in this context—including the burden, distress, and demands they encounter—deserves specific attention. Caregiver burden, defined as the multidimensional physical, psychological, and social strain associated with caregiving demands, has been extensively studied in chronic care settings and is associated with adverse caregiver outcomes [10, 11, 12]. However, these findings cannot be directly extrapolated to acute HaH. The onset of an acute HaH episode represents a distinct clinical context, characterized by an abrupt escalation of caregiving demands, often without prior preparation or adaptation period.

Existing evidence on caregiver experiences in HaH remains limited and methodologically heterogeneous. Several studies have focused on specific diagnostic populations, limiting the generalizability of their findings to broader acute care populations [13, 14, 15]. Comparative studies generally suggest that caregiver burden in HaH is, at worst, comparable to that observed in conventional hospitalization, with some studies reporting more favorable outcomes [16, 17]. However, qualitative and mixed‐methods studies highlight a more nuanced reality: while HaH is often perceived as more patient‐centered and responsive than hospital care, it may simultaneously increase the responsibilities placed on caregivers, sometimes to a challenging degree [18, 19]. Similarly, high caregiver satisfaction has been widely reported in HaH settings, but satisfaction and acceptability do not necessarily reflect low caregiver burden [20]. A recent scoping review further highlighted substantial variability across studies in caregiver outcome definitions, burden assessment tools, timing of evaluation, and HaH program models, limiting comparability and synthesis of findings [21].

Importantly, prior studies have rarely examined caregiver burden as a primary outcome and to our knowledge, have not specifically assessed it at the very onset of the acute HaH episode. This early phase is clinically critical: unrecognized high burden may compromise caregiver well‐being and sustainability of home‐based acute care. Psychosocial resources such as resilience and perceived social support are established protective factors in chronic caregiving [22, 23], but their role during acute care transitions remains unclear. Whether these resources similarly buffer caregiver strain at the acute onset of HaH—a context characterized by abrupt escalation of demands—was therefore examined as part of the present study.

The aims of this study were: (1) to assess the prevalence of high caregiver burden at the onset of the acute Hospital‐at‐Home episode, and (2) to identify caregiver characteristics associated with high burden at this early timepoint.

2. Methods

2.1. Study Design and Setting

We conducted a cross‐sectional observational study within the acute Hospital‐at‐Home (HaH) program of Clalit Health Services (CHS), the largest Israeli health fund, Northern District, Israel. Data collection took place between January 2023 and December 2024. This manuscript reports the baseline cross‐sectional analysis of a prospective observational cohort; a companion longitudinal analysis examining burden trajectories during the episode is being prepared for separate submission. The study was conducted in accordance with STROBE guidelines.

The CHS acute HaH program is a standalone service, delivering time‐limited (typically 4–7 days) hospital‐level care at home, including daily medical and nursing visits, intravenous therapy, and laboratory services. Escalation to conventional hospitalization occurs when clinically required.

Caregiver cooperation is a formal admission requirement; the designated caregiver is identified by the family and confirmed by the clinical team at the time of admission.

2.2. Participants and Recruitment

Eligible participants were caregivers of adult patients admitted to the CHS acute HaH program. Inclusion criteria were: (1) designated caregiver for a patient admitted during the study period, (2) age ≥ 18 years, (3) insured by CHS (required under the approved Helsinki Committee protocol) and (4) sufficient Hebrew proficiency to complete a telephone interview without an interpreter. Paid caregivers were not eligible.

The caregiver was defined as the person assuming primary responsibility for the patient during the HaH episode, as identified by the family and the clinical team. Where multiple family members were involved, the caregiver who assumed the greatest day‐to‐day responsibility was enrolled.

Recruitment was conducted in two stages. First, HaH nursing staff approached eligible caregivers and requested permission to share their contact details with the research team. In the second stage, the research team contacted caregivers by telephone, provided a full explanation of the study, and obtained verbal informed consent prior to any data collection. The recruitment flow is detailed in Figure 1.

FIGURE 1.

Flow diagram of caregiver recruitment, eligibility assessment, enrollment, and inclusion in the final analytic sample. Of 590 caregivers approached by Hospital‐at‐Home nursing staff, 217 (36.8%) agreed to be contacted by the research team. Following telephone contact, exclusions for ineligibility (non‐Clalit insurance, n = 24; insufficient Hebrew proficiency, n = 9), non‐completion of baseline assessment, withdrawal of consent, and patient rehospitalization or death resulted in a final analytic sample of 125 caregivers.

2.3. Data Collection

Structured telephone interviews were conducted by 3 trained research team members, independent of the clinical HaH team. Interviews were conducted within the first 48 h of the HaH episode. Only caregivers were interviewed. Patient data were obtained from CHS administrative records.

2.4. Measures

Caregiver Characteristics collected included age category, sex, marital status, education level, employment status, and socioeconomic status.

Patient Characteristics included age, sex, and the primary medical reason for admission to the acute HaH program.

Caregiving Context was assessed using study‐specific items documenting the caregiver's relationship to the patient, co‐residence, prior caregiving for the patient, whether this was the caregiver's first HaH episode as primary caregiver, and the availability of additional help at home.

Health‐related quality of life (HRQoL) of the caregivers was measured using the EQ‐5D‐5 L [24, 25]. Index scores were calculated using the UK value set, alongside the EQ‐VAS (0–100 scale). Official authorization was obtained from the EuroQol Group.

Psychosocial resources of the caregivers:

Resilience was assessed using the Connor–Davidson Resilience Scale—2 items (CD‐RISC‐2; range 0–8) [26]. High resilience was defined as a score of ≥ 7 consistent with the upper portion of the general population distribution [26]. An officially authorized Hebrew translation was used.

Perceived social support was assessed using a single unvalidated binary (yes/no) item asking whether the caregiver had someone to count on for help if needed. This approach is consistent with established single‐item support measures [27], though it does not capture the full multidimensional nature of social support.

Caregiver burden was assessed using the Caregiver Strain Index (CSI), a validated 13‐item instrument with strong psychometric properties and established clinical thresholds [28]. The CSI covers seven domains: employment impact, physical health, time demands, emotional adjustment, sleep disruption, personal life, and family relationships. Each item is scored binary (yes/no), yielding a total score of 0–13; a score of ≥ 7 defines high burden based on the original validation by Robinson (1983) [28]. The CSI was selected for its brevity and widespread use in caregiving research, validated Hebrew translation and suitability for telephone administration during an acute care transition. An officially authorized Hebrew translation was used.

2.5. Statistical Analysis

Analyses were performed using SAS version 9.4. Continuous variables are presented as mean ± standard deviation and categorical variables as frequencies and percentages. Percentages were calculated among participants with available data.

Caregivers with high burden (CSI ≥ 7) were compared with those with low burden using χ ² or Fisher's exact tests for categorical variables and independent‐samples t‐tests for continuous variables.

Multivariable logistic regression was used to examine factors associated with high caregiver burden. Candidate variables were selected based on clinical relevance and bivariate associations at p < 0.10. Socioeconomic status was excluded from the multivariable model because of a high proportion of missing data. A two‐step hierarchical model was specified to assess the stability of associations across clinically relevant domains. Step 1 included patient characteristics, caregiver characteristics, caregiving context variables, and psychosocial variables. Step 2 additionally included current employment status, admission diagnosis type, and first HaH episode as primary caregiver. Odds ratio (OR) with 95% confidence intervals (CI) is reported. Collinearity among EQ‐5D index, resilience, and perceived social support was assessed in model fitting and did not indicate problematic multicollinearity. Participants with missing values on any covariate included in the multivariable model were excluded using complete‐case analysis (N = 123).

Given the modest sample size and number of predictors, multivariable findings should be interpreted as exploratory and hypothesis‐generating. Statistical significance was set at p < 0.05.

3. Results

3.1. Participant Flow and Characteristics

Of 590 caregivers approached by HaH nursing staff, 217 (36.8%) agreed to be contacted by the research team. Of these, 125 caregivers (21.2% of those approached) provided informed consent and completed the baseline assessment, constituting the final analytic sample. The full recruitment flow is shown in Figure 1. The participation rate should be considered when interpreting the prevalence estimates reported below.

Baseline characteristics are summarized in Table 1. Patients had a mean age of 75.8 ± 12.6 years. The most common admission diagnoses were dehydration (25.6%) and urinary tract infection (20.0%). Caregivers were predominantly female (67.2%) and were most commonly the patient's spouse (45.6%) or adult child (46.4%). Overall, 79.2% had prior caregiving experience with the patient, and 64.8% were experiencing their first HaH episode as primary caregiver.

TABLE 1.

Baseline characteristics of patients and caregivers at the onset of the acute Hospital‐at‐Home Episode (N = 125).

Variable	Total (N = 125)
Patient characteristics
Age, years, mean ± SD	75.8 ± 12.6
Female sex	66 (52.8)
Type of acute condition
Acute illness	103 (82.4)
Acute exacerbation of chronic disease	22 (17.6)
Primary admission diagnosis
Dehydration	32 (25.6)
Urinary tract infection	25 (20.0)
Cellulitis	22 (17.6)
Pneumonia	17 (13.6)
Congestive heart failure exacerbation	16 (12.8)
COPD exacerbation	6 (4.8)
Other	7 (5.6)
Caregiver characteristics
Age category, years
< 40	13 (10.4)
40–60	42 (33.6)
60–70	35 (28.0)
≥ 70	35 (28.0)
Female sex	84 (67.2)
Married/partnered a	95 (77.9)
Academic education	58 (46.4)
Currently employed	56 (44.8)
Socioeconomic status ≤ average b	85 (81.7)
Relationship to patient
Spouse	57 (45.6)
Adult child/parent	58 (46.4)
Other	10 (8.0)
Co‐resident with patient (same household)	77 (61.6)
Prior caregiving for this patient	99 (79.2)
First HaH episode as primary caregiver	81 (64.8)
Caregiver health and psychosocial factors
Perceived social support (yes) d	86 (68.8)
High resilience (CD‐RISC‐2 ≥ 7; range 0–8)	78 (62.4)
EQ‐5D‐5L: any problems e
Mobility	39 (31.2)
Self‐care	16 (12.8)
Usual activities	40 (32.0)
Pain/discomfort	55 (44.0)
Anxiety/depression	63 (50.4)
EQ‐VAS (0–100), mean ± SD	76.5 ± 18.2
EQ‐5D index (UK value set), mean ± SD	0.79 ± 0.26
Caregiving context (among caregivers with prior experience, n = 99)
Duration of caregiving
< 6 months	16 (16.2)
6–12 months	12 (12.1)
1–2 years	15 (15.2)
2–5 years	22 (22.2)
5–10 years	14 (14.1)
> 10 years	20 (20.2)
Additional help at home c
None	32 (25.6)
Family member	57 (45.6)
Paid caregiver	34 (27.2)

Note: Values are n (%) unless otherwise specified. Percentages for variables with missing data are calculated among respondents with available data.

Abbreviations: CBS, Central Bureau of Statistics; CD‐RISC‐2, 2‐item Connor–Davidson Resilience Scale; COPD, chronic obstructive pulmonary disease; EQ‐5D, EuroQol 5‐Dimension questionnaire; EQ‐VAS, EuroQol visual analogue scale; HaH, Hospital‐at‐Home.

^a Married/partnered: 95 of 122 respondents (77.9%); 3 did not respond (missing).

^b Socioeconomic status assessed by comparison to the Israeli national average monthly salary (CBS 2021: 11,349 NIS/month). “≤ average” combines “below” and “similar” responses (85 of 104 respondents, 81.7%); 21 preferred not to answer (missing).

^c Additional help at home: 2 respondents selected “other” (missing). Percentages calculated among 123 respondents with available data.

^d Perceived social support assessed using a single binary (yes/no) item: “Do you have someone you can count on when you need help—for example, if you were ill yourself?”

^e “Any problems” defined as EQ‐5D‐5L level ≥ 2.

3.2. Prevalence of High Caregiver Burden

High caregiver burden, defined as a Caregiver Strain Index score ≥ 7, was observed in 61.6% of caregivers (n = 77) within the first 48 h of the HaH episode.

Burden differed according to prior caregiving experience (Supplemental Figure S1). Among the 99 caregivers with prior caregiving experience, 75.8% (n = 75) met the high‐burden threshold (CSI ≥ 7). Among the 26 caregiving‐naïve caregivers, 7.7% (n = 2) met the high‐burden threshold; both had prior HaH experience. Overall, 42.3% of caregiving‐naïve caregivers (n = 11) reported some burden (CSI > 0), with a mean CSI score of 3.6 among those with CSI > 0, whereas 57.7% (n = 15) reported no burden (CSI = 0). Among caregiving‐naïve individuals experiencing their first HaH episode (n = 22), none met the high‐burden threshold.

3.3. Psychosocial Factors and Health‐Related Quality of Life

Caregivers with high burden reported lower well‐being across several domains (Table 2). Compared with caregivers with low burden, those with high burden had lower perceived social support (59.7% vs. 83.3%, p = 0.006) and a lower prevalence of high resilience (52.0% vs. 79.2%, p = 0.002). They also more frequently reported problems with anxiety/depression (63.6% vs. 29.2%, p < 0.001) and usual activities (41.6% vs. 16.7%, p = 0.004) on the EQ‐5D‐5L. The mean EQ‐5D index score was lower in the high‐burden group than in the low‐burden group (0.74 ± 0.27 vs. 0.88 ± 0.23, p = 0.003). These findings are illustrated in Figure 2.

TABLE 2.

Patient and caregiver characteristics according to caregiver burden at the onset of acute Hospital‐at‐Home (CSI < 7 vs. CSI ≥ 7).

Variable	CSI < 7 (n = 48)	CSI ≥ 7 (n = 77)	p
Patient characteristics
Female sex	29 (60.4)	37 (48.1)	0.18
Age, years, mean ± SD	74.1 ± 12.6	76.8 ± 12.5	0.24
Admission type			0.79
Acute illness	39 (81.3)	64 (83.1)
Acute exacerbation of chronic disease	9 (18.8)	13 (16.9)
Caregiver characteristics
Female sex	29 (60.4)	55 (71.4)	0.20
Age category			0.63
Married/partnered a	38 (82.6)	57 (75.0)	0.33
Academic education	25 (52.1)	33 (42.9)	0.64
Currently employed	25 (52.1)	31 (40.3)	0.20
Socioeconomic status ≤ average b	30 (79.0)	55 (83.3)	0.58
Co‐resident with patient	29 (60.4)	48 (62.3)	0.83
First HaH caregiving episode	35 (72.9)	46 (59.7)	0.13
Caregiver health and psychosocial factors
Perceived social support (yes) c	40 (83.3)	46 (59.7)	0.006
High resilience (CD‐RISC‐2 ≥ 7)	38 (79.2)	40 (52.0)	0.002
EQ‐5D usual activities—any problems	8 (16.7)	32 (41.6)	0.004
EQ‐5D pain/discomfort—any problems	13 (27.1)	42 (54.6)	0.003
EQ‐5D anxiety/depression—any problems	14 (29.2)	49 (63.6)	< 0.001
EQ‐VAS, mean ± SD	75.0 ± 19.0	78.9 ± 16.7	0.25
EQ‐5D index, mean ± SD	0.88 ± 0.23	0.74 ± 0.27	0.003
Caregiving context
Prior caregiving for same patient	24 (50.0)	75 (97.4)	< 0.001
Duration of caregiving d			0.63
Additional help at home e			0.11

Note: High burden defined as CSI ≥ 7. EQ‐5D index calculated using UK value set. “Any problems” defined as EQ‐5D‐5L response level ≥ 2. p values from χ ² or Fisher exact test (categorical variables) and independent t‐test (continuous variables). (N = 125). Values are n (%) unless otherwise specified. Percentages for variables with missing data are calculated among respondents with available data.

Abbreviations: CD‐RISC‐2, 2‐item Connor–Davidson Resilience Scale; CSI, Caregiver Strain Index (range 0–13); HaH, Hospital‐at‐Home.

^a Married/partnered: Percentages calculated among respondents with available data (CSI < 7: n = 46, 2 missing; CSI ≥ 7: n = 76, 1 missing).

^b Socioeconomic status ≤ average: Percentages calculated among respondents with available data (CSI < 7: n = 38, 10 missing; CSI ≥ 7: n = 66, 11 missing). See Table 1 Footnote b for definition.

^c Perceived social support: Single binary (yes/no) item. See Table 1 footnote e for item wording.

^d Duration of caregiving analyzed only among caregivers with prior caregiving experience (n = 99). Distribution by category is presented in Table 1.

^e Additional help at home: Distribution by category (none, family member, and paid caregiver) is presented in Table 1. Only the overall p value for the comparison between burden groups is reported here.

FIGURE 2.

Caregiver Health‐Related Quality of Life, Psychosocial Characteristics, and Caregiving Context According to Caregiver Burden at the Onset of Acute Hospital‐at‐Home. Data are presented according to caregiver burden at the onset of the HaH episode, defined using the Caregiver Strain Index (CSI), with low burden (CSI < 7) and high burden (CSI ≥ 7). Panel A displays the proportion of caregivers with prior caregiving experience for the same patient, perceived social support, high resilience, and problems in EQ‐5D domains (anxiety/depression, pain/discomfort, and usual activities). Panel B shows mean EQ‐5D‐5L index scores (UK value set) with 95% confidence intervals. EQ‐5D outcomes reflect the presence of any reported problem (Levels 2–5). Resilience was assessed using the CD‐RISC‐2 (score ≥ 7). Percentages are based on available data. CSI, Caregiver Strain Index; EQ‐5D‐5L, EuroQol 5‐dimension 5‐level.

3.4. Factors Associated With High Caregiver Burden

In multivariable analysis (Table 3), perceived social support (OR 0.22, 95% CI 0.06–0.74, p = 0.01) and high resilience (OR 0.33, 95% CI 0.11–0.95, p = 0.04) were each associated with lower odds of high burden. Paid caregiver assistance was associated with higher odds of high burden (OR 5.23, 95% CI 1.17–23.37, p = 0.03). The direction and magnitude of these estimates were stable across Step 1 and Step 2 of the hierarchical model.

TABLE 3.

Multivariable associations with high caregiver burden at the onset of acute Hospital‐at‐Home (CSI ≥ 7).

Variable	Step 1 OR [95% CI]	p (step 1)	Step 2 OR [95% CI]	p (Step 2)
Patient female sex (vs. male)	1.94 [0.79–4.73]	0.15	1.98 [0.80–4.92]	0.14
Patient age (per year)	1.03 [0.98–1.07]	0.24	1.03 [0.99–1.07]	0.20
Caregiver female sex (vs. male)	0.88 [0.33–2.38]	0.80	0.92 [0.33–2.55]	0.87
Caregiver age a	0.82 [0.52–1.29]	0.39	0.78 [0.47–1.28]	0.32
Caregiving context b
Additional helper—family member (vs. none)	3.18 [0.90–11.26]	0.07	3.05 [0.86–10.91]	0.09
Additional helper—paid caregiver (vs. none)	5.36 [1.24–23.14]	0.03	5.23 [1.17–23.37]	0.03
Perceived social support (yes vs. no) c	0.21 [0.06–0.71]	0.1	0.22 [0.06–0.74]	0.01
High resilience (CD‐RISC‐2 ≥ 7; vs. < 7)	0.35 [0.12–1.00]	0.05	0.33 [0.11–0.95]	0.04
EQ‐5D index (per 0.1 unit increase)	0.16 [0.02–1.37]	0.09	0.22 [0.02–2.16]	0.19
Acute exacerbation of chronic disease (vs. acute illness)	—		0.67 [0.21–2.16]	0.51
Currently employed (yes vs. no)	—		0.73 [0.27–1.96]	0.53
First HaH caregiving episode (vs. prior experience)	—		0.62 [0.24–1.59]	0.32

Note: Step 1 included patient characteristics, caregiver characteristics, caregiving context, and psychosocial factors. Step 2 additionally included admission diagnosis type, current employment status, and first HaH caregiving episode. OR < 1 indicates lower odds (protective association); OR > 1 indicates higher odds of high burden. Hierarchical logistic regression (N = 123). Outcome: High caregiver burden (Caregiver Strain Index ≥ 7).

Abbreviations: CD‐RISC‐2, 2‐item Connor–Davidson Resilience Scale; CI, confidence interval. CSI, Caregiver Strain Index; EQ‐5D, EuroQol 5‐Dimension questionnaire; HaH, Hospital‐at‐Home; OR, odds ratio.

^a Caregiver age: Coded as an ordinal variable with four levels (< 40, 40–60, 60–70, ≥ 70); OR reflects change per one ordinal level increase.

^b Additional help at home: Two respondents selected “other” (missing). Percentages calculated among 123 respondents with available data.

^c Perceived social support: Single binary (yes/no) item. See Table 1 footnote e for item wording.

In bivariate analysis, prior caregiving experience was associated with high burden (p < 0.001). First HaH episode as primary caregiver was not significantly associated with high burden in bivariate analysis (p = 0.13) and did not reach significance in the multivariable model (OR 0.62, 95% CI 0.24–1.59, p = 0.32).

A cross‐tabulation of prior caregiving experience and first HaH episode status yielded four subgroups. Among experienced caregivers in their first HaH episode (n = 59), 13 had low burden and 46 had high burden. Among experienced caregivers with prior HaH experience (n = 40), 11 had low burden and 29 had high burden. Among caregiving‐naïve individuals in their first HaH episode (n = 22), all 22 had low burden and none had high burden. Among caregiving‐naïve individuals with prior HaH experience (n = 4), two had low burden and two had high burden. These findings are presented in Table S1.

Other factors, including patient age, patient sex, and admission diagnosis, were not significantly associated with high burden in the final multivariable model.

4. Discussion

In this real‐world study of acute HaH admissions, nearly two‐thirds of caregivers had high burden (CSI ≥ 7) within the first 48 h of the episode. This prevalence is higher than that reported in earlier HaH clinical trials [14, 29, 30], likely reflecting differences between controlled trial populations and the heterogeneous real‐world setting studied here. Because the CSI was administered at HaH onset without a formal pre‐admission time anchor, this burden should be interpreted as reflecting caregiver strain present during the early HaH transition, rather than burden attributable exclusively to the HaH episode. In a healthcare system where long‐term care support is often limited and fragmented, caregivers may assume substantial responsibilities outside the acute care system. Early identification of caregiver strain within the HaH admission workflow therefore represents both a clinical priority and a practical opportunity to recognize vulnerable caregivers and tailor support from the outset.

The near‐universal prevalence of prior caregiving among high‐burden caregivers highlights the cumulative nature of strain in this population. Prior caregiving experience is an exposure variable, not a validated measure of burden; its association with high CSI scores is consistent with accumulated strain but does not establish long‐standing exhaustion. The duration of prior caregiving was not a significant predictor, possibly reflecting adaptation over time or survivor bias among those who remained in the caregiving role.

The contrast between experienced and caregiving‐naïve caregivers is clinically instructive. Among experienced caregivers, high burden was highly prevalent; the CSI at onset reflects a composite of pre‐existing strain and early HaH‐related burden whose contributions cannot be disentangled in a cross‐sectional design. Among caregiving‐naïve individuals overall, high burden was uncommon, occurring in only two caregivers, both of whom had prior HaH experience.

Notably, among caregiving‐naïve individuals experiencing their first HaH episode, none met the high‐burden threshold, although a substantial proportion reported early low‐level burden. This suggests that abrupt assumption of a new caregiving role in a medically intensive home setting may generate measurable strain even when it does not yet reach the threshold for high burden. High burden was prevalent among experienced caregivers regardless of prior HaH experience, while caregiving‐naïve individuals in their first HaH episode had uniformly low burden. This pattern suggests that accumulated prior caregiving strain, rather than unfamiliarity with the HaH model itself, may be a central driver of burden at episode onset. This tentative pattern must be interpreted with extreme caution given the very small sample size of the naïve/prior HaH group (n = 4). Prospective longitudinal studies are needed to determine whether early low‐level burden among caregiving‐naïve individuals escalates during the acute episode.

Beyond caregiving history, psychosocial resources appeared clinically relevant at HaH onset. In multivariable analysis, perceived social support and high resilience were each associated with lower odds of high burden, consistent with the broader caregiving literature [31, 32]. Given the cross‐sectional design, reverse causality cannot be excluded: caregivers with lower burden may be more likely to report higher resilience and greater perceived support, rather than these resources necessarily conferring protection. The attenuation of the EQ‐5D index association after adjustment for resilience and social support is consistent with, but does not establish, partial mediation; formal mediation analysis would require a longitudinal design.

The association between perceived social support and caregiver burden should be interpreted with caution. The social support measure used in this study was a single binary item. Although conceptually grounded and methodologically supported [27], it has not been validated as a standalone caregiving instrument and does not capture the multidimensional nature of social support, including type, source, quality, and adequacy of support. More broadly, given the limited events‐per‐variable ratio and wide confidence intervals, all multivariable associations should be considered exploratory and hypothesis‐generating.

The association between paid caregiver assistance and high burden also requires cautious interpretation. Although paid caregiver assistance was associated with higher odds of high burden, the confidence interval was wide, reflecting limited precision in this subgroup.

This association most plausibly reflects the fact that families engage paid help precisely because patient complexity, functional dependence, or pre‐existing caregiver strain is already high—rather than paid assistance itself causing burden. This finding requires replication in larger samples with more detailed measures of patient functional status, care needs, and intensity of paid support.

4.1. Clinical Implications

These findings have direct implications for the organization of acute HaH admissions. Recent HaH implementation literature emphasizes that hospital‐level care at home requires complex local workflows, staffing models, and operational infrastructures, and that caregiver roles vary substantially across programs [33]. In this context, caregiver assessment should be viewed as part of HaH admission readiness, rather than as an optional patient‐experience measure. In practice, a brief assessment of prior caregiving experience, perceived social support, and emotional distress at admission could help identify caregivers most likely to benefit from early support, including social work involvement, anticipatory guidance, and closer follow‐up during the acute episode. Caregiving‐naïve individuals represent a group warranting proactive engagement, given that early low‐level burden was observed in this subgroup. The feasibility and effectiveness of such approaches in routine acute HaH care require prospective evaluation.

The need to define, support, and measure the caregiver role in HaH has been identified as a priority for future HaH research [3]. Our findings support this agenda by showing that caregiver vulnerability is already measurable at the point when hospital‐level care is initiated at home. They also reinforce the importance of measuring caregiver‐centered outcomes directly. Patient‐centered or system‐level outcomes, including time at home, satisfaction, or avoidance of hospitalization, should not be assumed to reflect caregiver well‐being. Future studies should evaluate whether structured caregiver screening and targeted support can be feasibly embedded into routine HaH workflows and whether such interventions improve caregiver and patient‐centered outcomes.

This study has several strengths. It was conducted in a real‐world, heterogeneous acute HaH population, with no exclusion by admission diagnosis, enhancing external validity. Data were collected at a precisely defined and clinically meaningful timepoint: The onset of the acute episode. Validated instruments with officially authorized Hebrew translations were used throughout. To our knowledge, this is one of the first studies to examine caregiver burden as a primary outcome in acute HaH, and the first to assess it specifically at episode onset. The multidimensional characterization of caregiver vulnerability, including resilience, health‐related quality of life, perceived social support, and caregiving context variables, represents an additional strength.

Several limitations should be acknowledged. First, the cross‐sectional design precludes causal inference, and reverse causality cannot be excluded. Second, the participation rate raises the possibility of selection bias. Recruitment depended on HaH nursing staff during acute admissions, and operational barriers may have limited the pool of caregivers approached. Because non‐approached individuals and nonparticipants were not characterized, the direction of this bias cannot be determined. Third, the multivariable model was limited by the sample size and events‐per‐variable ratio, and adjusted associations should therefore be interpreted as exploratory. Fourth, patient functional status, frailty, cognitive impairment, and delirium were not collected, limiting assessment of patient‐level contributors to caregiver burden. Fifth, perceived social support was assessed using a single binary item rather than a multidimensional validated instrument. Sixth, the Hebrew proficiency requirement and single‐program design may limit generalizability to linguistically diverse caregivers and to HaH programs with different caregiver expectations, staffing models, and in‐home support structures.

5. Conclusion

This study identifies caregiver burden as a clinically relevant issue at the onset of acute Hospital‐at‐Home care. Caregiver vulnerability appears to reflect not only the acute HaH episode, but also the caregiving context and psychosocial resources present when hospital‐level care is initiated at home. These findings support incorporating brief caregiver assessment into HaH admission workflows to identify individuals who may benefit from early support. Future prospective studies should evaluate caregiver burden trajectories during HaH and test whether targeted support interventions improve caregiver‐centered outcomes.

Author Contributions

Shira Duhamel: study concept and design; acquisition of data; analysis and interpretation of data; drafting of the manuscript. Sophia Eilat‐Tsanani: study concept and design; critical revision of the manuscript for important intellectual content. Maya Lerner‐Shikory: study concept and design; acquisition of data. Adel Iktelat: study concept and design; acquisition of data. Alexander Lustman: study concept and design. Ron Sabar: study concept and design. Eti Zarbiv: acquisition of data.

Funding

This work was supported by Israel association of family physicians (IMA).

Disclosure

This study was supported by a research grant from the Israel Association of Family Physicians through the Israel Medical Association (IMA). The sponsor had no role in the study design, methods, participant recruitment, data collection, and data analysis, interpretation of results, manuscript preparation, or decision to submit the manuscript for publication.

Consent

Written informed consent was obtained from all individuals named below: Shiraz Vered, MSc—Statistical analysis; Roi Bitton—Interviewer; Ita Tamarkin, RN—Nurse and interviewer.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Distribution of caregiver burden at the onset of the acute Hospital‐at‐Home episode according to prior caregiving experience. Stacked bar chart showing the proportion of caregivers with high burden (CSI ≥ 7), low burden (CSI 1–6), and no burden (CSI = 0) among caregivers with prior caregiving experience (n = 99) and caregiving‐naïve caregivers (n = 26). Among experienced caregivers, 76% met the high‐burden threshold and 24% had low burden (CSI 1–6); none had CSI = 0. Among caregiving‐naïve caregivers, none met the high‐burden threshold; 42% (n = 11) had low burden (mean CSI 3.6) and 58% (n = 15) had no burden (CSI = 0). For statistical comparisons, see Table 2. CSI = Caregiver Strain Index; HaH = Hospital‐at‐Home.

Table S1: Cross‐tabulation of caregiver burden according to prior caregiving experience and first Hospital‐at‐Home (HaH) episode status.