Introduction

Currently, most critically ill patients with COVID-19 are receiving mechanical ventilation (MV) worldwide. Studies have estimated that 6%– 37% of patients who require MV in intensive care units (ICUs) eventually require long-term mechanical ventilation (LTMV) [1, 2]. Therefore, the number of ventilator-dependent patients is rapidly increasing worldwide. The most widely used definition of LTMV is the continuous use of MV for at least six hours per day over a period of more than 21 days [3].

The prevalence of LTMV varies among countries. In Europe, Canada, and the US, 7–15 people per 100,000 require LTMV [3–6]. In Taiwan, the rate is considerably higher, at 26 people per 100,000 [7]. To more effectively manage ventilator-dependent patients and control medical expenses for LTMV [8], the National Health Insurance Administration (NHIA) of Taiwan has established an integrated delivery system (IDS) with a comprehensive prospective payment program for LTMV patients [8]. This system comprises integrated payments with a managed care system that covers mechanical ventilators in four phases of care and payment schemes: a fee-for-services payment in ICUs covers patients who require acute respiratory care who stay in the ICU for up to 21 days. After 42 days of per-diem payment, if MV is still needed, patients are transferred to a respiratory care center (RCC) or to subacute care for aggressive weaning of ventilator support for LTMV patients. After 64 days of unsuccessful liberation from MV, patients who become ventilator-dependent are transferred to a per-diem respiratory care ward (RCW) and/or per-diem home care services [8, 9]. An RCW provides 24-hour skilled nursing for ventilator-dependent patients who need long-term care. The home care services support patients who are cared for directly by family caregivers at home by providing additional care from a multidisciplinary team [10].

To facilitate improved home care services, it is crucial to implement policies that shift resources toward community-based care for LTMV patients. Currently, most such patients and their families show a preference for RCW institutional care services. According to statistics published in 2011 by the Taiwanese Ministry of Health and Welfare, only 8.1% of LTMV patients select home care services after being discharged from the hospital, in contrast to patients in Canada and the US, of whom 11.0–30.8% choose home care services [11, 12].

To investigate the factors determining health care service choice, Gehlbach and Salamanca [13] conducted a cohort study involving 548 MV patients in the ICU. The researchers found that patient age, duration of MV use, cognitive dysfunction, and functional status were all associated with the decision to be discharged from a care facility. There is a large body of research that examines the role of patient-related [13] economic and environmental factors [14]. However, the role of resource factors in determining health care service choices for LTMV patients is less well understood. Andersen’s Behavioral Model (ABM) proposes that predisposing, enabling, and need factors [15] have been widely used to explore the determinant factors of health care service use. There is little research using ABM as a theoretical framework for exploring the determinants of LTMV patients’ health care choices. The present study seeks to fill this gap in the literature by using the ABM to examine influencing factors of health care service choices among LTMV patients.

Methods

Results

The age of the study population was 66.67 ± 20.41 (mean ± SD) years, and slightly more than half of the participants were male (53.4%) and non-single (56.4%) (Table 1). Nearly one-third (30.7%) had an elementary school-level education only, and 69.6% were members of a nuclear family. Finally, 52.3% had low SES. The patients had 1.91 ± 0.92 family caregivers available to provide care (Table 1). These caregivers demonstrated a moderately strong level of belief (3.49 ± 0.61) in providing care to the patients, and the patients perceived a moderate degree of social support from both their health care providers (2.22 ± 0.64) and their family members (2.20 ± 0.67). Almost half of the patients resided in highly urbanized areas (46.8%), and their annual household income was USD 27,681.7 ± 28,159.5. Over three-quarters (78.9%) of the patients did not receive government subsidies. The patients exhibited 1.98 ± 1.21 chronic diseases, and the duration of ventilator use was 3.56 ± 2.51 years (Table 1). Over half of the patients (56.4%) had moderate to severe cognitive impairment. Finally, a large proportion (80.5%) were severely or totally dependent on assistance for activities of daily living (ADL).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of long-term mechanical ventilation patients (N = 365).

https://doi.org/10.1371/journal.pone.0274303.t001

Regarding LTMV service choice, a total of 197 patients (54.0%) had chosen to utilize the RCW service, with the remaining 168 (46.0%) receiving home care services (Table 1). According to the bivariate analysis of the predisposing factors, we found that patients who were non-single (t = 3.78, p = 0.047) and lived with extended family (χ² = 12.25, p = 0.002) were significantly more likely to choose home care services.

Regarding enabling factors, we found that stronger caregiving beliefs among family caregivers (t = −7.92, p < 0.001) and greater social support from health care providers (t = −3.13, p = 0.002) and family (t = −3.23, p = 0.001) were significantly positively associated with the selection of home care service. However, patients with more caregivers (t = 3.96, p < 0.001) were more likely to choose RCW services. In terms of need factors, we found that fewer years of ventilator use (t = −1.98, p = 0.048), moderate to severe cognitive impairment (χ² = 95.53, p < 0.001), and severe to total dependence on assistance for daily functions (χ² = 88.25, p < 0.001) were positively associated with the use of RCW services. Regarding the factors that affect health care service choice, in the first hierarchical logistic regression model (Model 1) for the factors influencing the selection of RCW or home care service, which included only predisposing factors, we found that older patients (odds ratio [OR] = 0.97, 95% confidence interval [CI] = 0.96–0.98) were more likely to choose RCW services, while patients who were non-single (OR = 1.85, 95% CI = 1.11–3.86) chose home care services 1.85 times more often than single patients. Patients living with their nuclear (OR = 2.54, 95% CI = 1.06–6.06) or extended families (OR = 3.01, 95% CI = 1.15–7.90) were more likely to choose home care services than patients who lived alone (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Hierarchical logistic regressions of utilization among patients with long-term ventilator dependence.

https://doi.org/10.1371/journal.pone.0274303.t002

In Model 2, which included both the predisposing and enabling factors, we found that only age and marital status remained significant. In addition, patients with higher numbers of family caregivers (OR = 0.65, 95% CI = 0.48–0.87) were more likely to choose RCW services, and those with caregivers who held stronger caregiving beliefs (OR = 1.09, 95% CI = 1.06–1.12) tended to choose home care services. A greater level of perceived social support from health care providers (OR = 1.68, 95% CI = 1.05–2.66) was also more likely to lead to the selection of home care services.

In Model 3, all three factors (predisposing, enabling, and need) were included (Table 2). Of the predisposing factors, we found that older patients were more likely to choose RCW services, and patients who had completed at least junior high school were more likely than illiterate patients to choose RCW services. By contrast, non-single patients and those residing with extended family were more likely to choose home care services than patients who lived alone. Patients with a high or medium SES were more likely to choose home care services than patients with a low SES.

Regarding the enabling factors, the same variables that were included in Model 2 were included in Model 3. Patients with higher numbers of family caregivers were again more likely to choose RCW services, whereas those whose family caregivers held stronger caregiving beliefs and those who felt they received more social support from their health care providers were more likely to choose home care services. Of the need factors, we included the cognitive function and daily functions in Model 3. Patients with normal cognitive function (OR = 7.87, 95% CI = 3.72–16.64) or mild cognitive impairment (OR = 3.64, 95% CI = 1.62–10.58) were more likely to choose home care services than those with moderate cognitive impairment. Patients with a moderate level of dependence for daily functions (OR = 7.57, 95% CI = 1.84–31.27) were more likely to choose home care services than those with severe dependence. We calculated the Cox-Snell R² value for Model 3 and found that the model accounted for 44.8% of the variance in long-term health care service utilization (Table 2).

Discussion

We used the ABM to identify factors that affect health care service utilization by LTMV patients. The model explained 44.8% of the variance in their choice of long-term health care service. In our sample, 54.0% of LTMV patients chose to reside in RCWs, while the remaining 46.0% relied on home care services. This was similar to the findings of a cohort study by Hill and Fowler [2], in which 43.5% of the patients selected home care services; however, our study indicated a lower percentage of choosing home care services than that of reported by Gehlbach and Salamanca [13], which was 73.9%. One possible explanation could be related to out-of-pocket health expenditures. In Taiwan, the health insurance fully covers LTMV patients’ health care services at RCWs or home care services, patients or their families only pay low out-of-pocket costs. In the United States, if individuals decide to use institutional services, the amount of services is decided by health providers [36] results in higher out-of-pocket expenditures than home care services [37]. Hence, there appears to be a higher probability of patients selecting home care services.

We found that older LTMV patients were more likely to choose an institutional care model. This finding is similar to results elsewhere in the Gehlbach and Salamanca study [13]. A possible explanation is that older patients have more severe illnesses and chronic conditions [13, 38] and may therefore be in greater need of professional health care services and may rely more heavily upon institutional care. Patients with higher education levels were more likely to choose institutional services. Our findings are similar to those of Tzeng [39], who indicated that patients with higher education were more likely to choose institutional care. However, this finding contrasts with those of Wu and Hu [25] and Kim and Cho [26], who found that education level was not associated with long-term health care utilization. The reason for these differences may be the distribution of education levels within the study samples. In our study, 51.0% of the participants had completed junior high school and therefore had received at least seven years of education. By contrast, in Wu and Hu [25] sample, only 22.8% of the long-term elderly patients in their sample had completed junior high school, while only 38.4% of the stroke patients in the study by Kim and Cho [26] had more than six years of education. A higher level of education is associated with a higher SES, probably because of which a more educated patient is more likely to choose professional or institutional health care services [40].

Regarding family structure, patients living with extended family were more likely to choose home care services. Our findings are supported by Zhang and Zeng [40], who reported that patients living with a spouse, children, and extended family members may enjoy greater access to family resources [40]. Elsewhere, evidence suggests that residing with extended family engenders a greater number of relationships with other family members and stronger attitudes of thoughtfulness among members [41]. It is reasonable to expect patients living with extended family to be more likely to choose home care services.

Differences in the SES may reflect differences in social class, and an individual with a higher SES is assumed to have a greater sense of mastery and control and may be more likely to select the best health care service available [42]. High-SES families have more resources available with which to facilitate a patient’s transition from institutional care to home care. In addition, caregivers may exhibit greater role readiness and adaptability in high-SES families than in low-SES families [42]. Thus, it is understandable that a higher SES is a determinant of the selection of home care services among LTMV patients. Our findings regarding sex are concordant with those obtained elsewhere, which suggest that sex does not significantly affect health care service choice [13, 25, 26, 38, 40].

Of the enabling factors investigated, patients with more caregivers were more likely to choose RCW services. This finding is similar to that of Lee [43] but dissimilar from those of a number of other studies [25, 40, 41]. One possible explanation may be related to the severity levels of disability. LTMV patients with a higher severity level of disability require 24-hour daily care and total assistance and caregivers competent in skills, such as nasogastric feeding, inhalation, suction, oxygen cannula, and saturation monitoring, as well as in ventilator techniques. In the present study, of the LTMV patients who received RCW services, 98.5% had severe to total dependence for daily functions, higher than in case of those who received home care services (59.5%). In addition, among the home care patients in this study, 80 (47.6%) of their families hired people for additional assistance for patient care and life assistance at home, often facilitating patients receive 24-hour care for patients [44]. However, significant difference was observed in the availability of the number of family caregivers between RCW and home care services. Thus, taking into consideration the availability of additional people in home care services, it may not be sufficient to differentiate between use of RCW and home care services. Future studies should examine the relationship of the number of caregivers and their competence with health care service choice for LTMV patients.

This is the first study to examine the relationship between family caregiver beliefs surrounding the provision of care and the choice of health care service. We found that the beliefs of family caregivers influenced the health care service choices of LTMV patients. The measures used in this study reflect the attitudes of the primary caregiver toward the patient [45]. Family caregivers with stronger beliefs about the importance of providing care are likely to exhibit a stronger commitment to patient care [46, 47]. Such caregiver beliefs can therefore motivate patients to choose to receive health care at home rather than in an RCW setting [48]. Therefore, stronger family caregiver beliefs in the importance of providing care are potentially linked to a stronger commitment to caregiving by family members and may in turn make patients more likely to choose home care services.

The social support offered by health care providers was identified as a significant factor determining health care service choice. When LTMV patients require relocation or a change in health care service type, professional advice and relevant resources can greatly reduce their psychological distress and that of their family members if there are conflicts involved in this change [49, 50]. Furthermore, support from health care providers helps patients and family members cope with the realities of receiving and providing care at home [50]. Greater levels of support from health care providers can also facilitate adaptation to both the patient and caregiver roles [50, 51], and formal support from health care providers can facilitate improved adherence among patients and promote effective health care behaviors [51, 52]. Therefore, when patients perceive greater levels of social support from health care providers, they are understandably more willing to choose home care services.

Social support from the family was not found to be associated with health care service choice. One possible explanation for this finding relates to psychological burden. In a study by Liu and Lu [53], the majority of family caregivers of LTMV patients experienced a heavy psychological burden, regardless of whether the patient was receiving institutional care or home care. This burden may explain the lower levels of social support offered by family members than by health care providers.

Another enabling factor, urbanization, did not significantly affect health care service choice. This finding is similar to that of Pai and Kung [14] but dissimilar to findings of Wu and Hu [25] and Zhang and Zeng [40]. These differences in research findings may be attributable to differences in the study setting. Our study uses a sample drawn from North Taiwan, a region where resources are allocated relatively uniformly across areas; therefore, the level of urbanization may not have varied sufficiently to significantly affect health care service choice. Household income also did not exert a significant effect on health care service choice, which is similar to the findings of Luppa and Luck [22] and Wu and Hu [25]. The receipt of government subsidies also exerted no significant effect. In Taiwan, the government provides welfare subsidies for long-term care equipment, such as respiratory therapy equipment, wheelchairs, and pressure air mattresses. All LMTV patients and family members who receive welfare subsidies select either RCW or home care services with equally likelihood. Therefore, government subsidies may not significantly affect health care service choices.

Regarding need factors, we found that cognitive function and ADL had significant effects on health care service choice. Not surprisingly, LTMV patients with normal or mild cognitive impairment were 7.87 and 3.64 times more likely to choose home care services, respectively. Our findings are similar to those obtained by other studies [13, 22, 26]. Cognitive impairment is an aging-related disease [54], and elderly patients with more chronic diseases also tend to exhibit more cognitive function decline [13, 38]. Severe cognitive impairment is a strong influencing factor of institutional care choice. The degree of disability also has a significant influence on patients’ institutional care choice. Our findings are similar to those obtained by other researchers [13, 22, 26, 40]. Patients with more severe cognitive impairment are more limited in their capacity to perform ADL and therefore require more support and assistance. Thus, older patients who are highly dependent on assistance for ADL are more likely to choose institutional care.

Neither the number of chronic diseases nor the number of years of ventilator use significantly affected health care service choice. Our findings are supported by results obtained by other researchers [13, 55]. It is worth noting, however, that the within-sample variation in the number of chronic diseases and the duration of ventilator use in this study were low, which limited the ability of the regression analysis to identify the relationship between these variables and health care service choice.

A limitation of our study is the use of a retrospective cross-sectional analysis design for enrolled LTMV patients receiving home or institutional care. In reality, the choice of health care service for LTMV patients may be a dynamic process. Therefore, capturing a longer-term picture of the health care service choices of such patients is likely to provide a more comprehensive understanding. In the future, we recommend that a longitudinal prospective study be conducted to obtain a more accurate understanding of the influencing factors of health care service choices of LTMV patients, specifically how such choices evolve over time. In addition, based on our finding that family caregivers play an important role in the health care service choices of LTMV patients, we recommend that future studies focus more on caregiver-related factors, such as caregiver burden.

Conclusion

To our knowledge, the present study is the first to use the ABM to identify factors that affect health care service choices among LTMV patients. Our model offers a number of valuable insights into the determinants of the health care service choice of LTMV patients from both theoretical and practical perspectives. Our model accounts for 44.8% of the observed variance in health care service choice. Evidently, there is more to the question of health care service choice that has yet to be explored. Our study indicates that patients who are older and have higher education levels, more caregivers, more severe cognitive impairment, and a higher level of dependence are more likely to choose RCW services. By contrast, patients who are not single, reside with extended family, receive more social support from health care providers, have a higher SES, and have caregivers who express stronger beliefs in the importance of providing care are more likely to choose home care services. It is therefore imperative that more efforts be made to reinforce primary caregivers’ beliefs in the importance of providing care. It is also important to increase the level of social support offered by health care providers so that patients and their family members are better positioned to choose the most appropriate health care service. Future research should explore the cost-effectiveness or cost utility of different health care service models.

Supporting information

Acknowledgments

The authors would like to thank all those who participated in the study.