Preferences for home and hospital treatment under different conditions. HHA indicates home health aide.
Fried TR, van Doorn C, O'Leary JR, Tinetti ME, Drickamer MA. Older Persons' Preferences for Home vs Hospital Care in the Treatment of Acute Illness. Arch Intern Med. 2000;160(10):1501–1506. doi:10.1001/archinte.160.10.1501
Copyright 2000 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2000
Although the home is expanding as a potential site for acute illness treatment, little is known about patients' preferences for home vs the hospital.
To determine older persons' preferences for home or hospital as a treatment site for acute illness and factors associated with preference.
Two hundred forty-six community-dwelling persons aged 65 years or older hospitalized with congestive heart failure, chronic obstructive pulmonary disease, or pneumonia were identified in 2 urban teaching hospitals and received telephone interviews 2 months after hospitalization. They were asked their preference for home or hospital treatment, given the availability of equivalent therapies and outcomes at the 2 sites and a nursing visit and several hours of home health aide assistance daily in the home. They were also asked about changes in preference with changes in the description of the outcome or the availability of services.
If home and hospital offered equivalent outcomes, 46% of the sample preferred treatment at home. Preferences were heavily dependent on the outcome of the illness, physician opinion about the best site of care, and the provision of house calls. Higher education, white race, living with a spouse, being deeply religious, and having 2 or more dependencies in activities of daily living were associated with a preference for home treatment.
Under conditions of equivalent outcome, preferences for treatment site are almost equally divided between home and hospital. Explicit elucidation of preferences is necessary if patients' preferences are to play a meaningful role in decision making about site of care.
THE ROLE of home care has been dramatically changing during the past several decades. Originally considered a substitute for institutional care for patients with long-term care needs, the advent of home intravenous therapy, oxygen, and other treatment modalities has resulted in the use of home as an early discharge site for hospitalized patients. More recently, acute care in the home is being considered as a complete substitute for hospitalization. This movement is in part a response to concerns about the iatrogenic complications associated with hospitalization of frail older persons and the hypothesis that home care may avoid these complications.1,2 Another powerful incentive shifting acute care into the home is economics. Medicare's prospective payment system gave an initial boost to the provision of "high-tech" care outside the hospital,3 and the substitution of home for hospital is expected to result in further cost savings. On the other hand, many physicians are skeptical about the safety and efficacy of home care,4 and they identify a lack of ancillary supports as a major barrier to caring for patients with acute illness at home.5
Despite the skepticism about the feasibility of treating acute illness in the home, a growing body of literature indicates that conditions previously thought to require hospitalization can safely and effectively be treated at home. Home treatment is as effective as hospital treatment of deep venous thrombosis,6 the proportion of patients with community-acquired pneumonia treated as outpatients can safely be increased,7 and a home hospital program has been shown to decrease the need for hospitalization compared with historical controls.8 Explicit criteria have been developed to prospectively identify older patients with congestive heart failure, chronic obstructive pulmonary disease, and pneumonia who are currently routinely admitted to the hospital but who could safely be treated at home,2 and efforts to increase home treatment of these patients is under way.9
Although much work has been done to demonstrate the feasibility of home care in the treatment of acute illness, little is known about patients' preferences for treatment site. The few studies of how decisions regarding treatment site are made indicate that patient preference for treatment site is rarely elicited and even more rarely incorporated into decisions about site of care.5,10 The different forces both encouraging and discouraging the further development of acute care in the home highlight the importance of evaluating patient preferences. Such an evaluation would provide an understanding of the value to patients of having a choice in treatment site. While patient preferences should ideally be at least one consideration in all treatment decisions, in those patients for whom hospital and home provide similar outcomes, preference should be the primary determinant of treatment site.11
Older persons' preferences for treatment site are particularly important for several reasons. First, patients aged 65 years and older make up a consistently increasing proportion of hospital utilization, accounting for 38% of discharges and 48% of the total days of inpatient care in 1995.12 Second, home care for patients who are rendered homebound by their illness is a Medicare benefit, thus potentially facilitating its more widespread use. Finally, older patients have been targeted for expanded home care programs in an effort to avoid the high incidence of functional decline associated with their hospitalization.2 The purpose of this study was to describe preferences for treatment site among older persons with pneumonia, congestive heart failure, and exacerbation of chronic obstructive pulmonary disease, conditions identified as potentially amenable to treatment in both the home and the hospital.
Participants were consecutive community-dwelling persons aged 65 years and older who were hospitalized with a primary diagnosis of congestive heart failure, chronic obstructive pulmonary disease, and/or pneumonia in 2 urban teaching hospitals between July 1, 1997, and January 31, 1998. We chose to study recently hospitalized patients because an earlier study conducted among patients receiving home care services failed to identify anyone treated for these diseases in their home.13 Potential participants were excluded if they had extreme hearing loss; did not speak English; were too demented to participate in a telephone interview; did not have a telephone; lived out of state; were discharged to long-term nursing home care, to inpatient hospice care, or to another hospital; were participating in other ongoing longitudinal studies involving frequent telephone follow-up; their physician failed to provide consent for the participant to be contacted about the study; or they were in the hospital at the time of the interview. This study was approved by the Human Investigations Committee of the Yale University School of Medicine, New Haven, Conn.
Of the 415 potentially eligible participants, 15 could not be reached. Of the remaining 400 subjects, 246 agreed to participate, for a response rate of 62%. Among those who were eligible, nonparticipants were significantly older than participants (78.2 vs 75.6 years; P<.001) and were less likely to have a diagnosis of chronic obstructive pulmonary disease (17% vs 26%; P=.02). A greater proportion of men than women were nonparticipants (45% vs 37%; P=.07). There were no differences by diagnosis of pneumonia or congestive heart failure between respondents and nonrespondents.
Participants were interviewed by telephone 2 months after the hospitalization to minimize the burden of the study.
Participants were asked to indicate their preference for home or hospital as site of care under an initial baseline state and to choose from a list the primary reason for their preference. They were then asked if their preference changed in response to changes in the baseline state. The baseline state, list of reasons for preference, and changes in the baseline state were all derived from previous qualitative research.13
To have the preference be rooted in actual experience rather than a hypothetical scenario, participants were asked to reflect back on the illness that led to their hospitalization. Because the qualitative research demonstrated that the probable outcome of the illness, the availability of different treatments, and the availability of services in the home all influence preference, these factors were all initially constrained by specifying a baseline state in which (1) home and hospital would provide an equal likelihood of survival; (2) the same treatments, such as intravenous medications, oxygen, blood tests, and x-rays, would be available in the home and in the hospital; and (3) a daily nursing visit and several hours of home health aide assistance would be provided at no cost to the patient. This preference question has a test-retest reliability of 0.54, measured by the κ coefficient.
All participants were then asked to choose from a list the most important reason underlying their preference. For home, the choices included (1) poor hospital experience, (2) greater comfort, (3) ability to do things when participant wished, (4) not being around other sick people, (5) having family around, (6) fear of not returning home from the hospital, and (7) other. For hospital, the choices included (1) feeling of greater safety, (2) ease on family, (3) constant availability of help with tasks such as going to the bathroom, and (4) other.
Each of the variables of the baseline state was then changed, and the participant was asked if this resulted in a change in his or her preference (Figure 1). Participants were asked to think about the impact of each of the changing variables on their originally stated preference. The change in variables differed depending on the initial preference. Participants initially preferring the hospital were asked their preference if (1) a nurse visited several times each day, (2) a nurse was in the home during the nights, (3) a nurse was in the home for the first 24 hours of the illness, (4) the physician made house calls, (5) the physician believed that home would be the better treatment site, or (6) home treatment would be less likely to leave them in a weakened state after the illness than hospital care. Participants initially preferring home were asked their preference if (1) the physician believed that the hospital would be the better treatment site, (2) they could not get to the bathroom by themselves, (3) they could not get out of bed themselves, or (4) there was a greater chance for recovery with hospital treatment than with home treatment. Participants who initially preferred the hospital were not asked their preference if there was a greater chance for recovery with home treatment because the previous qualitative research suggested that they would find this impossible to believe.13
Independent variables, representing factors potentially associated with preference for treatment site,13 fell into the following categories: perceptions of home and hospital as treatment sites, sociodemographic variables, health-related variables, social support variables, and a measure of attitudes toward health. Perceptions of home and hospital as treatment sites consisted of a series of questions asking respondents to agree or disagree with different aspects of acute care as delivered in the home and hospital, including the comfort of care, rapidity of recovery, safety, quality of care, and burden on friends and family at home and in the hospital.
Sociodemographic variables included age, race, education, and report of sufficiency of income, taken from the Established Populations for the Epidemiologic Study of the Elderly (EPESE).14 Health-related variables included the diagnosis responsible for hospital admission, comorbidities self-rated health, functional status as measured by the Activities of Daily Living Scale15 and the Instrumental Activities of Daily Living Scale,16 cognitive status as measured by the Folstein Mini-Mental State Examination,17 and depression as measured by the Geriatric Depression Scale.18
Social support variables included the participant's living arrangements, number of children and proximity to the participant's household, whether the participant had someone who could help during an illness, and whether the participant felt that more help was needed during an illness. Participants were also asked their religion, how religious they considered themselves to be, and to what extent religion was a strength and comfort to them.
Participants' attitudes toward their health were assessed by means of the Multidimensional Health Locus of Control Scale, which consists of 3 subscales measuring the degree to which people agree that they have control over their health (internal subscale), that their health is due to chance (chance subscale), and that others (physician, family) have an influence on their health (powerful others subscale).19 For the present study, the original response format of a 5-point Likert scale ranging from "strongly disagree" to "strongly agree" was simplified to a 3-point scale, "agree," "neither agree nor disagree," and "disagree," scored as 1, 2, and 3, respectively. Valid responses could range from 6 to 18, with higher scores indicating a higher level of agreement.
Univariate statistics were used to describe the participant population and the frequency of their preferences for home or hospital as a treatment site. The χ2 and Fisher exact tests for categorical variables and t tests for continuous variables were used to identify factors associated with preference. For categorical variables, the relative risk was used to describe the strength of the association between the factor and the preference. In the case of perceptions of home and hospital, the relative risk expresses the likelihood of preferring home given agreement with the item as compared with either disagreeing or being neutral about the item.
Characteristics of the 246 participants are shown in Table 1. When asked about their treatment preference site under the baseline conditions of similar therapies being available in the home and in the hospital with a similar expected outcome, and having a daily nursing visit and several hours of home health aide assistance, 113 participants (46%) preferred treatment at home, and 132 (54%) preferred treatment in the hospital, with 1 respondent being unable to decide about preferred treatment site.
The primary reason cited by the participants who preferred home was greater comfort by 61 (54%), the presence of family by 22 (20%), a bad hospital experience by 15 (13%), the ability to do things when participant wished by 9 (8%), and a variety of other considerations by 6 (5%). Of the 15 who had a bad hospital experience, lack of nursing response and/or a sense that the nurse did not care was the experience cited by 6 of the participants. Complications believed to be a result of the hospitalization were cited by 4 participants, being bothered by personnel in training and having a bad experience with roommates were each cited by 2 participants, and 1 participant cited a bad experience in the emergency department. The primary reason cited by the participants who preferred the hospital was the sense of greater safety in the hospital by 84 (64%), the constant presence of help for tasks such as getting to the bathroom by 31 (23%), greater ease on family by 10 (8%), and a variety of other considerations by 7 (5%).
Changing the conditions of the illness dramatically reduced participants' preference for treatment at home, as shown in Figure 1. Only 9% of participants who initially preferred home would continue to do so if the hospital were associated with a greater chance of getting better, and only 12% would prefer home if their physician believed that the hospital would be better. Almost one half of the 113 who initially preferred the home instead preferred the hospital if they could not get out of bed unassisted. On the other hand, the majority of participants who initially preferred hospital care continued to prefer the hospital, regardless of the conditions of the illness or the services provided. The service most likely to result in a change in preference from hospital to home was the provision of physician house calls, in which case 24% of those initially preferring the hospital would prefer home. The only condition resulting in a substantive shift in preference was if the physician believed that home would be the better site, in which case 48% of participants who initially preferred the hospital instead preferred home.
As shown in Table 2, negative perceptions about hospital care and positive perceptions about home care were associated with a preference for treatment in the home. Supporting the finding that comfort was the most commonly cited reason for preferring home, the belief that home is more comfortable than the hospital was most strongly associated with a preference for home treatment. Conversely, negative perceptions about home care and positive perceptions about hospital care were inversely associated with a preference for treatment in the home. The belief that hospital is more comfortable than home, that people recover faster in the hospital, and that home treatment would be burdensome on friends and family were most strongly inversely associated with a preference for home treatment.
Higher socioeconomic status, living with a spouse, and being deeply religious were all associated with a preference for treatment at home. The magnitudes of these associations were modest (all relative risks were less than 1.42) (Table 3). Despite the small numbers of severely functionally dependent patients, this characteristic was strongly associated with preference, as 69% of these patients preferred treatment at home. Although white race also appeared to be associated with a preference for home treatment, this finding failed to reach statistical significance, presumably because of the small number of African American participants. Of the 3 subscales of the Multidimensional Health Locus of Control Scale, only the powerful others subscale was associated with preference. Those preferring home scored, on average, 1 point lower on this subscale, indicating less belief that other people have control over one's health. Primary diagnosis, comorbidities, self-rated health, cognitive status, depression, Instrumental Activities of Daily Living status, having children, and requiring more assistance when ill all showed no association with preference for treatment site.
Older persons' preferences for site of treatment in acute illness are variable and highly dependent on the probable outcome of treatment, the opinion of the physician, and the services provided. Although preferences are strongly associated with perceptions of home and hospital, they are not, with the exception of functional status, strongly associated with sociodemographic or health status factors. Preferences frequently changed depending on the circumstances of the illness and the services provided. This mutability of preference is reflected in the only moderate test-retest reliability of the preference question. From discussion with participants during pilot testing, this did not appear to result from difficulty with or lack of clarity of the question, but rather from changes in preference that resulted from the participant having given further thought to the question after the initial interview.
Preferences for home were more easily swayed than were preferences for the hospital, suggesting that the concept of treatment at home is new and unfamiliar to many patients. The extreme responsiveness of preferences to physician opinion and the provision of physician house calls demonstrates that physicians' attitudes toward and practice of home care may be the most important determinant of patients' preferences for treatment site. This notion is supported by the findings of another study of older hospitalized patients' preferences for treatment site. In this study, which described a program in which the patient would receive a physician house call daily, 72% of patients said they would prefer treatment at home.20
These findings have clear implications for decision making about acute illness if preferences are to inform these decisions. Home treatment is a desired alternative only if it provides equivalent outcomes to hospitalization. This contradicts the notion that, for many older persons, "comfort" is more important than "cure," or, in other words, that quality-of-life considerations override quantity of life.21,22 These results are similar to the findings of previous studies demonstrating the willingness of older persons to undergo intensive care unit treatment23 and their unwillingness to trade off any amount of time in their current health for perfect health.24
However, in the case of equivalent outcomes, the preference expressed by many of the respondents for treatment at home supports its explicit consideration and expansion as a treatment alternative for less severely ill patients. The lack of a strong association between preference and sociodemographic and health status implies that the only way to ascertain these preferences is to ask the patient directly. If home and hospital can offer equivalent outcomes, then patient preference should be an explicit component in the decision about where to treat.
On the other hand, the strongly held preference for the hospital of many participants indicates that many people want to be in the hospital when they are acutely ill. Even if home were associated with a better functional outcome, 74% of those who initially preferred the hospital under conditions of equivalent outcome continued to do so. This finding is especially important in light of efforts to increase the use of home treatment for acute illness to achieve cost savings.7 How preferences should be weighed against competing economic considerations is a question under active debate. In the case of alternative treatments offering equivalent outcomes but at different costs, the weight to be given to preference is particularly unclear. It could be argued that, analogous to performing magnetic resonance imaging in the evaluation of a tension headache, hospital care for low-severity illness is not medically indicated and, therefore, patient preference is irrelevant.25 Unlike magnetic resonance imaging, a relatively new technology never routinely incorporated into the evaluation of all headaches, the hospital has, until recently, been considered the appropriate treatment site for all but those with the most trivial illnesses. Patients' preferences for hospitalization, accompanied by the reasonable expectation that the hospital is the "right" place to receive treatment, cannot be easily dismissed.
The study has several limitations. The 62% participation rate and differences between participants and nonparticipants suggest that the study population may not be representative of the population of older hospitalized patients. However, because participants and nonparticipants differed according to factors that were not associated with preference, the implications for the generalizability of the results are unclear. In addition, by interviewing patients who had survived their hospitalization and had returned home, we studied patients with "successful" hospitalizations who may therefore have more favorable attitudes toward the hospital. However, they represent the majority of older patients with acute illness, as few have experienced home care for their acute illnesses. This study also points to the importance of additional research regarding preferences. Although we believe that the only marginal test-retest reliability of the single preference question used in this study reflects real change in the attitudes of the participants, it is possible that it instead reflects a weakness in the question itself. If this is the case, then it may be possible to measure preference for treatment site more reliably. Finally, the mutability of preferences suggests that they may be influenced by clinical experiences and may therefore be highly variable over time. Further study is necessary to determine the stability of preferences over time and the potential for changes in preference with different clinical experiences.
The associations of perceptions of home and hospital with preference for treatment site illustrate the importance of certain aspects of care, regardless of treatment site. The reasons chosen by respondents for their preferences show that comfort and safety are the 2 greatest concerns of older patients regarding their site of care. The associations of perceptions of home and hospital with preference for treatment site also illustrate the importance of concern about burden on families and friends. The example of the birthing room provides a good analogy as to how the process of hospital care can be redesigned to incorporate patient-valued aspects of care, bringing the "comforts of home" to the delivery room.26 To the extent that treatment either at the hospital or at home responds to patients' concerns about safety, comfort, and placing a burden on patients' friends and families, the site of care may be less important than the aspects of care of which it is composed.
Accepted for publication October 18, 1999.
This work was supported by the Picker/Commonwealth Clinical Scholars Program, the Oxford Foundation, and the Paul Beeson Physician Faculty Scholars Program, New York, NY, and the Claude D. Pepper Older Americans Independence Center of Yale University, New Haven, Conn.
We thank Risa Lavizzo-Mourey, MD, MBA, for her instructive comments on the design of the interview instrument and presentation of the results and Elizabeth Bradley, PhD, for her thoughtful review of the manuscript.
Corresponding author: Terri R. Fried, MD, Geriatrics and Extended Care 240, Veterans Affairs Connecticut Healthcare System, 950 Campbell Ave, West Haven, CT 06516 (e-mail: email@example.com).