Factors Influencing Nursing-Sensitive Outcomes in Taiwanese Nursing Homes

The search for nursing-sensitive clinical outcomes has been hampered by the lack of patient care sites that truly reflect nursing interventions and changes in health status upon which nursing care has had a direct influence. Recent changes in health care policy in Taiwan, ROC created an unexpected and unique opportunity to explore factors related to effective nursing practice and its impact on measurable clinical outcomes. The increase in number of nurse-managed nursing homes offer health care settings where patient care outcomes are likely to be sensitive to the effects of nursing care. This paper reports the results of research conducted in eight selected nursing homes in Taiwan which was designed to explore factors influencing biopsychosocial functioning and patient satisfaction as nursing-sensitive outcomes. The researchers operationalized the concepts of biopsychosocial function and patient satisfaction with care as outcomes and considered the impact of individual patient and organizational factors on those care outcomes. The following key hypotheses were generated and tested:

1. Individual and organizational factors have a positive effect on patients' biopsychosocial functioning.
2. Patient satisfaction with care contributes to biopsychosocial functioning.
3. Nursing practice patterns have a positive effect on biopsychosocial functioning and an indirect positive effect through patient satisfaction with care.
4. Patients' physical, psychological, and social functioning are positively related.

Background

As a result of the rapid socioeconomic changes in Taiwan, the family as an institution is undergoing major change. As the family changes, so too does its ability to continue to meet traditional roles and responsibilities. The changing family structure is reducing the ability of the family to provide home care for its aging family members. The provision of care outside the family in long-term care facilities is emerging as an important part of enabling older people to live in the community with dignity. To meet the needs of the changing society, the National Assembly in 1992 amended the Constitution, stipulating that the province of Taiwan should implement a national health insurance and promote research and development of modern and traditional medicines (Department of Health, 1993). This legislation strengthened the 1991 Nursing Law of the Republic of China (ROC) which was promulgated with provisions for care for chronically ill older patients in nursing care institutions that provide rehabilitation as well as physical care.

Because of these changes in health policy, the Department of Health (DOH) plans to shorten the length of hospital stays by providing for long-term care of discharged or chronically ill patients in the community. Limited medical care resources can then be used more effectively for patients in acute need. Since 1991, the DOH has been strengthening the following rehabilitation and long-term care programs through:

1. establishing a medical care delivery system for chronic diseases and rehabilitation;
2. setting up nursing homes on a trial basis;
3. implementing community care and home health care;
4. training more professional workers in rehabilitation and for the care of chronic patients to be added to the staffs of medical care institutions; and
5. improving and expanding facilities for rehabilitation and care of chronic patients in medical care institutions of all levels.

The philosophic shift from curing disease to long-term care has created an opportunity for professional nursing to examine and evaluate its practice.

 

The first nursing home was set up in 1991 on a pilot project basis by the DOH. The majority of patients were older people with cardiovascular or cerebrovascular disease, diabetes mellitus, senile dementia, and osteoregression. Most of the patients were referred by their families, and needed assistance with activities of daily living. The average length of stay in 1995 was 55.8 days with a low turnover rate. Although the National Health Insurance program was implemented in March, 1995, standards for establishing nursing homes have not as yet been approved. Most nursing interventions in Taiwan are based on hospital practices that have rarely been evaluated for their effectiveness or efficacy.

Although based on research completed in 1995, this paper contributes new ideas to the discussions of nursing classification by Gordon (1998) and Clark (1998) in the Sept. 30 issue of the Online Journal of Issues in Nursing by focusing on specific nursing-sensitive patient outcomes -- biopsychosocial functioning and satisfaction with care. Clark (1998) supports the need for standardized languages and classifications for nursing around the world, noting that translation of languages developed in and for the USA may not be relevant in other countries. Gordon (1998) cites national and international efforts to develop a nursing language system that includes diagnoses, interventions and outcomes. The culturally developed instruments used in the presented research could be assistive to other professional nurses interested in the development of an international classification system.

Conceptual Framework

The conceptual framework for organizing this research was developed based on a search of the patient outcomes literature and an adaptation of the conceptual model developed by Munson, Beckman, Clinton, Kever, and Simms (1980) and supported by Beckman and Simms (1992) and Simms (1993). In their guide to redesigning nursing practice patterns, Beckman and Simms (1992) provided field-tested instruments to measure elements of any nursing practice pattern in acute care settings: patient characteristics, nursing resources, work characteristics and organizational support. Beckman and Simms (1992) thus provide a useful framework for investigating nursing-sensitive patient outcomes. Although designed to investigate the influence of nursing practice patterns on patient and organizational outcomes, the model supported the importance of individual patient and organizational characteristics in any practice pattern redesign and outcomes research. Figure 1 portrays the relationship between individual and organizational factors and their effect on biopsychosocial functioning as the unique multidimensional outcome variable believed to best reflect nursing care in settings where nursing personnel are the predominant care givers. Patient satisfaction was conceptualized as a mediating variable, based on the work of Donabedian (1980), who has consistently regarded patient satisfaction as an outcome. He believes that satisfaction with care represents the patient's judgment of quality of nursing care. Since perception is subjective, satisfaction was considerd a mediating variable in this study.

Figure 1: Conceptual Framework for Factors Influencing Nursing-Sensitive Outcomes

Related Literature

In a review of the literature on long-term care in Taiwan, nursing care was not mentioned as a relevant key word before 1992. Acknowledgment of the importance of long-term care was mentioned in reports during the governmental-wide programs in 1991, while the nursing research on long-term care was mentioned only during and after the passage of the Nursing Law. Therefore, in conducting this investigation, the researchers reviewed diverse literature outside of Taiwan on topics which yielded insight into individual and organizational factors considered to have an impact on biopsychosocial functioning and patient satisfaction as nursing-sensitive outcomes.

Outcome Measures

Historically, outcomes of health care have been measured through mortality rates, morbidity, disability, unscheduled readmission, unscheduled repeat surgery, or the extent and severity of pathology as measured by laboratory reports and pathologists (Stewart & Ware, 1992). Outcome measures are primarily medically oriented assessments of patient care outcomes. From a medical perspective, patient outcomes are defined in terms of palliation, control of illness, cure, or rehabilitation (Lohr, 1988). Physicians thus base a successful outcome almost exclusively on the disappearance of symptoms without considering that patients often have to live with a variety of residual health and functional limitations afterwards (Defriese, 1992).

Until the 1970s, most clinical studies defined outcome as a dichotomous variable in terms of death or life (Grady & Swartz, 1993). Unscheduled readmission has also been recognized as an essential outcome indicator. Rehospitalizations for adverse events following medical procedures have also been studied as part of appropriateness of care investigations (Riley, Lubitz, Gornick, Mantnech, & Egger, 1993; Chassin, Kosecoff, Park, Winslow, & Kahn, 1987; McClellan & Brook, 1992). Other outcome assessments such as the Patient Outcomes Research Team (PORT) projects, which are designed to evaluate the outcomes and costs of alternative services or procedures for the prevention, diagnosis, treatment, and management of specific clinical conditions, have assessed alternative types of surgery, nonsurgical interventions, and diagnostic tests of various diseases (AHCPR, 1991). In general, the focus of outcomes research has been disease oriented, which can lead to fragmentation of care, and the exclusion of consideration of psychological and sociological factors which influence health (Holden, 1991).

Physiological status is also frequently discussed in the patient outcomes literature and includes the measurement of processes that maintain life (Levine, Bramwell, Pritchard, Perrault, & Findlay, 1993). Evidence from research into the non-physiological areas of the links between depression and mortality (Vogt, Pope, Mullooly, & Hollis, 1994), culture and physical health (Wu, 1990), and income and health status (Kraus & Liang, 1993), implies that approaches to health care could benefit from widening the parameters of study and intervention from the purely physiological. As awareness grows as to the number of those suffering from long-term and chronic health care problems, the aim of curing becomes less central to health care endeavors (Lawton, 1991; Reed & Watson, 1994).

Patient Satisfaction

"Outcomes of Care" is an ambiguous concept and it is difficult to approach a consensus definition in the outcomes research (Brett, 1989). "Nursing-Sensitive Outcomes" is even more ambiguous. Thus patient satisfaction is increasingly considered an outcome in its own right (Lohr, 1988) and can be viewed as an objective of care representing the patient's judgment of quality care (Donabedian, 1980). It provides the information on the success of meeting the patient's values and expectations on which the patient is the ultimate authority. Expressions of satisfaction with elements of care can be considered non-intrusive indicators of the art and skill of nursing practice.

Moreover, improved technology yielding a choice of treatments, increased competition among health providers, and increased consumer awareness have resulted in the need to include patient satisfaction and quality of life as outcome dimensions (Grady & Schwartz, 1993). Ware Jr.'s (1991) study of the patient's perspective on outcomes of medical care combines traditional measures of disease status with the patient's view of disease and treatment. He concludes that it is impossible to move from efficacy research to effectiveness research unless new instruments to define effectiveness include quality of life, function and well-being from the patient's perspective.

Organizational Factors

A few researchers have attempted to develop instruments through the study of nursing practice patterns and their effect on quality patient care. Autonomous nursing practice has an effect on nurses' satisfaction (Beckman & Simms, 1992) which is frequently cited as a desirable organizational outcome that has a reciprocal relationship with patient satisfaction (O'Connor & Gibson, 1990). Hughes and Marcantonio (1992) examined the practice patterns of 155 staff nurses from selected home health, public health and hospital nurses using a quantitative perspective. They found that home health nurses have more opportunities to use their best clinical skills than do hospital nurses because of the less restrictive practice environment. Reed and Watson's (1994) qualitative study of the ways in which nurses assess the mobility of elderly patients, indicates that in long-term care settings where cure is clearly not an option for patients, the medical model is not consistent with nursing values, and may have a negative effect on the development of alternative approaches to care.

For many of the reviewed studies on individual and organizational contexts, analysis is based on large data sets, gathered for different purposes than for the particular outcomes research study (Colantonio, Kasl, Ostfeld, & Berkman, 1993; Sugisawa, Liang, & Liu, 1994; Wolinsky, Callahan, Fitzgerald, & Johnson, 1993). Data sources are population survey (Krause & Liang, 1993) or governmental health care projects evaluation (Miller, 1987), such as Medicare and Medicaid (Office of Nursing Home Affairs, 1975). It is noted thus that potential sources of error should be considered when ensuring the quality of these data and ascertaining the findings. Cherkin (1992) also argues that although literature which initially synthesizes and analyzes existing databases may be most appropriate, the most convincing scientific evidence for or against a specific intervention will require experimental designs such as randomized controlled trials. Hegyvary (1991) further indicates that the database for assessing outcomes should be extensive, including clinical, functional, financial and perceptual aspects.

In the studies reviewed, appropriate use of the techniques for group analysis were limited. None of those cited examined this aggregate data for its reliability, validity, and adequacy at the aggregated level. This may result in ecological fallacies of interpretation and generalization of results (Ferketich & Verran, 1992). Since correlation of aggregated variables based on homogeneous groups are higher than their individual-level counterparts (Ferketich & Verran, 1992), the dangers of using aggregated data to study individuals should be considered seriously.

In sum, the study of nursing-sensitive outcomes as measured by changes in biopsychosocial function and patient satisfaction in nurse-managed long term care settings is a neglected areas of research. In many of the studies reviewed for this investigation, analysis was based on large data sets gathered for different purposes. The nursing literature in Taiwan does not focus on patients in long-term care or on patient care outcomes. Current patient care, however, requires more than medical technology to achieve optimal outcomes and the role of the nurses becomes increasingly important especially in institutions where the predominance of care is managed and delivered by nursing staff. Evidence from the public media and general information indicates that the majority of institutionalized nursing home patients in Taiwan are sick and dependent on others for activities of daily living, and they tend to have limited physiological, functional, and social capabilities. Thus, this study was an attempt to seek empirical evidence of outcomes of care which are sensitive to nursing interventions. Recognizing the limitations of studying only demographic and other individual factors led the researchers to consider the impact of both individual patient and organizational factors.

Method

This study was conducted in eight currently established skilled nursing homes which represent various types of nursing homes in three main geographic areas in Taiwan. A purposive sample was drawn from the homes which were selected based on their differing characteristics. The target population consisted of patients who

1. had been living in the selected nursing homes for at least two weeks and
2. were mentally competent to participate in the study.

After obtaining informed consent, a three-part self- (N=55, 24.6%) or interviewer-administered (N=169, 75.4%) questionnaire was used with 224 patients who met the above selection criteria by the researchers. Nursing managers and clinical staff who were responsible for nursing care delivery determined patients' competency to participate in the study following an orientation to the project and consistent instructions for selection based on the following criteria: oriented to person and place, able to speak, and having the ability to respond to the cognitive items on the questionnaire. Of the 241 originally selected participants, 17 were excluded because of their cognitive orientation scores or failure to complete the questionnaire. Eight nursing home executives completed a questionnaire that provided data about organizational characteristics.

 

Instruments

Three major instruments were used to collect data for this study, namely, the Biopsychosocial Functioning Scale containing six subscales, the Nursing Organization Characteristics Questionnaire, and a two-part Patient Satisfaction Questionnaire. Additionally individual patient data were collected. Biopsychosocial functioning was measured by the Biopsychosocial Functioning Scale (BSF) which was adapted from both the Sickness Impact Profile (SIP) (Moinpour, McCorkle, & Saunders, 1988; Gerety, Cornell, Mulrow, Tuley, Hazuda, et al., 1994) and the Health Status Questionnaire (HSQ) (Ware Jr., 1992, 1991; Stewart and Ware, 1992; InterStudy, 1991; Stewart, 1992). The BFS is composed of a 28-item Likert type scale assessing six specific health dimensions: cognitive orientation, health perceptions, social interaction, mental health, bodily pain, and physical health. Three cognitive items that measured the orientation to people, place, and time were excluded from the total score because of the selection process for inclusion in the study. The total score for the BFS thus was calculated by summing the remaining 25 items of the BFS.

Health perceptions, psychological (measured by mental health) and social functioning (measured by social interaction), and bodily pain were all measured by a 5-point Likert scale. Nine items were selected to assess physical functioning, a dimension measuring the degree of limitations and capacities, mobility, self-care, and vision and hearing, as well as continence. Physical functioning was measured by a 3-point Likert scale which ranged from "limited a lot" to "not limited at all." Overall, the BFS is culturally relevant for this particular nursing home population in Taiwan. The response choices and wording were modified from both the SIP and HSQ to better capture specific limitations of interest, to describe more accurately the scale level defined by each item, and to facilitate verbal administration of the tool.

The Individual Patient's Data Tool included demographic characteristics, patient's previous experience with nursing care, and current health condition. The tool was designed by the researchers. Experience with nursing care was covered in 13 items about the patient's perception of activities of daily living and nutrition care. Data about the subject's health conditions (dependency level, medical care and rehabilitative interventions, mental conditions and behavioral conditions) refer to the severity of illness and were collected from clinic staff who were delivering care to patients participating in the study.

The Nursing Organization Characteristics Questionnaire consists of 11 questions pertaining to the qualifications of their nursing staff, practice patterns, and admission criteria. These data were needed to explore and compare the impact of nursing organizational factors on patient's biopsychosocial functioning.

Patient satisfaction with care, although qualitative and subjective in nature, was measured quantitatively in the study using two scaling procedures: magnitude estimation and dimension rating. Both procedures require judgments of intensity or strength of perception. Dimension rating was used to measure the strength and direction of patients' subjective perception about

1. technical skills,
2. interpersonal skills, and
3. the amenities of the care setting.

Magnitude estimation procedures avoid the use of a preset rating dimension (Wills & Moore, 1994). Instructions were given that emphasize the subjects' responses should be proportional to the intensity of experienced subjective states. This response was obtained through a single-item question rating quality of care on a scale ranging from 0 to 100.

 

When the focus of the study is specific to one culture, a symmetrical translation which is faithful to meaning done by someone with equal familiarity and colloquialism in each language, is required (Jones and Kay, 1992). Since the present study was conducted in Taiwan and the subjects were all Chinese, several strategies were used to avoid the threats to validity and reliability of the BFS and all scales used in the study. It included a back translation, a comprehensive literature review in the field, a panel of four experts and bilingual interpreters, and a pilot testing of the instrument with a convenience sample of eight nursing home patients. Overall, the Chinese version of the instrument was treated as a new instrument; reliability of the scales used in the study all fell within an acceptable range. Internal reliability coefficients range from 0.73 to 0.92 (Table 1, below).

 

Table 1. Reliability coefficients of study scales
Scale Biopsychosocial Functioning Health Perception Physical Functioning Psychological Functioning Social Functioning Cognitive Functioning Satisfaction Scale Experience With Care	Number of items 25 4 9 5 6 3 7 8	Alpha 0.80 0.85 0.82 0.73 0.77 0.79 0.92 0.92

 

Construct validity was assessed and established by convergent and discriminate validity. The intra-scale correlation among the BFS scale and its subscale items indicated that they belong together as measures of the overall construct/concept, thus establishing construct validity. Correlation, however, was generally modest to moderate, indicating that the subscales are not all measuring the same thing, an indication of discriminate validity. The comparison between admission data and the day studied data was used to ascertain convergent validity of the BFS scale. In addition, the inter-scale correlation between the BFS and other scales used in this study further explored empirical validity.

 

Data Analysis

Individual Characteristics

Sociodemographic characteristics are described in Table 2, below:

 

Table 2. Sociodemographic characteristics of study population.
Variables	N	%
Sex
Male Female	154 70	68.40 31.10
Native Place
Taiwan Hakkas Mainland	97 10 117	43.30 4.50 52.20
Marital Status
Single Married Separated Divorced Widowed/Widower	78 19 38 9 79	34.80 8.50 17.00 4.00 35.30
Religion
None Folk belief Buddhist Taoist Christian Catholic	84 22 93 3 16 6	37.50 9.80 41.50 1.30 7.10 2.70
Education
Illiterate Literate, No formal schooling Elementary Middle school High School College	103 49 33 16 13 10	46.00 21.90 14.70 7.10 5.80 4.50
Occupation
Governmental employee/Teacher Labor Business Military/Police Farmer No fixed office hour Housekeeping None	13 41 13 77 22 9 44 5	5.80 18.30 5.80 34.40 9.80 4.00 19.60 2.20
Sources of income (marked all that applied)
Pension Employment or business Interest from saving Parents Children Relatives/Friends Social security	82 3 20 4 64 6 69	36.60 1.30 8.90 1.80 28.60 2.70 30.80
Health insurance
None Government employee/dependents Labor Military/dependents Veterans/dependents Medicaid Life insurance with medical service	24 23 33 1 98 42 2	10.70 10.30 14.70 0.40 43.80 18.80 0.90
Major care giver before admission
None Spouse Children Parents friends/relatives Nurse-aides/maids	104 18 69 6 23 8	46.40 8.00 30.80 2.70 10.20 1.80
Reasons for institutionalization
No family members Family members can not take care Family members are not willing to care	111 103 6	49.60 46.00 2.70

There are 70 females (31.1%) and 154 males (68.4%) ranging in age from 26 to 101 with a mean age of 73 (SD=10.4). The length of stay of subjects ranged from around 3 weeks to 14 years with the mean length of stay being 659 days (SD=843.5), most suffered from cerebral vascular disease (32%) or multiple senile comorbid diseases (21%).

The subjects' health conditions on the day studied were further rated by nursing staff or through patients' records. Substantial progress was observed during the stay in nursing homes on 7 of the 8 subscales: eating, transferring, dressing, and toileting functioning, mental and behavioral conditions, and decreased medical care. Net positive changes in health conditions between admission (M=24.82, SD=3.32, Mode=27) and the day studied (M=25.34, SD=2.8, Mode=27) were found, except where there were increases in rehabilitative interventions between the two time periods. The full range of the scale was 8 - 30; Scores on admission ranged from 13 - 30, while scores on the day of study ranged from 16 to 30.

Health perceptions of the subjects tended to be mildly negative, while their biopsychosocial function as well as other dimensions of functioning tended to be positive (Table 3, below).

 

Table 3. Biopsychosocial functioning of study population
Dimensions/items	M	SD	Adjusted M	Range
Health perception/ 4	11.85	3.18	2.96	4-20
Physical/ 9	35.16	8.11	3.91	13-45
Psychological/ 5	15.39	3.30	3.08	7-24
Social/ 6	19.60	3.73	3.27	11-29
Bodily Pain/1	3.81	1.12	3.81	1- 5
Psychosocial/ 15	46.84	7.29	3.12	28-67
Biopsychosocial/ 25	85.81	11.71	3.43	49-114

Note: Higher scores indicate more positive affection.

The study sample also tended to be satisfied with the care received (M=4.80, SD=0.92 along the 6-point Likert-type scale (See Table 4, below). Scores which were measured by the satisfaction scale, however, were not significantly correlated with the subjects' biopsychosocial functioning or the overall quality of care score which is an additional satisfaction score measuring the overall favorable attitude toward the care received in these settings.

Table 4. Satisfaction with care of study population
Items	M	SD
1. The amount of time and care you receive.	4.95	0.84
2. The support and understanding you receive.	4.90	0.85
3. The amount you talking things over with nursing staff.	4.72	0.97
4. The manners and attitudes that nursing staff treat you.	4.96	0.83
5. Your confidence to the professional ability of nursing staff.	4.92	0.80
6. Do you feel generally relaxed and free of tension in this environment?	4.61	1.09
7. Overall, how satisfied are you with your institutional life?	4.59	1.07
Total score of satisfaction scale	4.80	0.92
Overall quality of care (ranging from 0 to 100)	83.22	9.58

Note: Higher scores indicate higher degree of satisfaction.

Organizational Characteristics

The majority of the eight participating settings are private (N=6, 75%), evenly distributed around Taiwan (north:south:west-central=2:3:3) and average around 50 beds (ranging from 5-200 beds, see Table 5, below).

 

Table 5. Organizational Characteristics
Nursing home code	Ownership Type				N. of beds (available/ equipped)	N. of RN/NA	Practice pattern
	public	private	free standing	hospital -based
1 2 3 4 5 6 7 8	x x	x x x x x x	x     x	x x x x x x	5/30 23/50 25/50 200/442 20/42 80/260 30/42 50/50	1/2 6/11 3/10 37/49 2/14 16/24 3/10 4/8	functional modular modular functional team functional functional modular
Total (%)	2(25)	6(75)	2(25)	6(75)

Most of the basic nursing care is delivered by nurse-aides with less than one year of nursing experience (N=101, 79%) and minimal training in nursing. All registered nurses are female, having an education level of junior college or vocational nursing school and the majority (65.3%) having more than 5 years of nursing experience. The average ratio of patient/RN was 6 to 1, and 3.4 to 1 for nurse-aides. There was no clear pattern of care delivery among the study settings. Nevertheless, similar nursing practice patterns were observed: nurses provide skilled nursing techniques, rehabilitation, health education, and are in charge of the unit; nurse-aides provide the vast majority of basic nursing care and the assistance with subjects' activities of daily living which are labor-intensive care; nurse-aides tend to value minimizing time for care more than striving for improved patient function. Functional nursing is the norm during the night shift while a combination of functional, team, primary, or care management is used during the day shifts.

Results of Hypotheses Testing

Hierarchical multiple regression analysis was performed to test the strength of relationships between study variables.

Hypothesis 1: Individual and organizational factors have a positive effect on patients' biopsychosocial functioning. (supported)

The bivariate are reported first. Among the individual factors, gender (F=5.33, df=1/222, p<.05) and the number of insurance sources (F=4.31, df=2/221, p<.05) were significantly correlated to biopsychosocial functioning. Males (M=73.66) had slightly higher scores on biopsychosocial functioning than females (M=70.69). The mean score of biopsychosocial functioning for individuals with two insurance sources (M=82) are higher than those with one source (M=72.85), and much higher than those who had no insurance source (M=55.50). Among the organizational factors, only the type of organization is significantly related to subjects' physical functioning (F=5.87, df=1/222, p<.05). Hospital-based nursing homes had significantly higher mean scores (M=22.54) than did freestanding nursing homes (M=21.16) for subjects' physical functioning.

Multiple regression analysis was performed to further determine the variance each individual and organizational predictor contributed to the subjects' biopsychosocial functioning. Of the 32 individual and organizational factors considered, 8 of them failed to meet the entry tolerance level criteria for inclusion. The total 24 variables in the equation accounted for 33% of the variance (R=.57, R²=.33, F=4.04, p<.001) in biopsychosocial functioning.

Hypothesis 2: Patient satisfaction with care contributes to biopsychosocial functioning. (supported)

Satisfaction with care significantly contributed to patients' biopsychosocial functioning when all individual and organizational factors were controlled, accounting for an additional 6.5% (R=.63, R²=.39, R² change=.065, F=4.9, p<.001) of the variance. By adding the satisfaction measures, a total of 39% of the variance can be explained by the model on which this is based. It appears that satisfaction may be more than an intervening variable- perhaps another nursing-sensitive outcome. Satisfaction with care also contributed significantly to patients' biopsychosocial functioning, directly accounting for 4.5% of the variance (R=.21, R²=.04, F=4.3, p<.05).

Hypothesis 3: Nursing practice patterns have a positive effect on biopsychosocial functioning and an indirect positive effect through patient satisfaction with care. (not supported)

Nursing practice patterns were measured by seven variables: the way that nursing care is delivered, number of RN, nurse-aides, physician and other professionals involved, and the frequencies of physicians' and other professionals' visits. Nursing practice patterns increased the explained variance to 32% (R=.57, R²=.32, R² change=5.3%, F=3.0, p<.01) when individual factors were entered into hierarchical regression equations first. The increase in the variance explained, however, was not statistically significant (F change=1.70, p>.05). Nursing practice patterns did not significantly contribute to subjects' biopsychosocial functioning when individual and organizational characteristics were controlled. Two aspects of nursing practice patterns, the way care is delivered and the number of physicians involved, contributed indirectly to biopsychosocial functioning via satisfaction with care measured by an overall quality of care score (R=.31, R²=.09, R² change=.022, F=2.22, p<.05). Nevertheless, nursing practice patterns alone did not contribute significantly to biopsychosocial functioning (F=1.74, p>.05).

Hypothesis 4: Patients' physical, psychological, and social functioning are positively related. (supported)

Subjects' physical and social functioning (r=.19, p<.01) and their psychological and social functioning (r=.22, p<.01) are positively and significantly related whereas no relationship between physical and psychological functioning was detected in this study. Data from the patient's self-rated physical functioning were further analyzed by Pearson's correlation to compare with the staff-rated patient's health conditions or dependency levels. The results show a high intercorrelation between physical functioning and health conditions on admission (r=.53, p<.01) as well as on the day studied (r=.63, p<.01).

Discussion and Conclusions

This study yielded five major findings:

1. only individual factors significantly predict patients' biopsychosocial functioning, whereas the organizational factors are generally not significant;
2. the quality of care score and satisfaction score significantly predict subjects' biopsychosocial functioning although these two scores are not correlated;
3. nursing practice patterns do not predict subjects' biopsychosocial functioning, but they can indirectly affect it via satisfaction with care;
4. subjects' social and psychological functioning, and their social and physical functioning are positively and significantly related, yet there is no significant relationship between physical and psychological functioning; and
5. positive changes occur in subjects' health conditions during their stay in study settings regardless of the nursing practice patterns.

The major reason for institutionalization of the elderly is the level of access to family care and social relationships. Those in better health (as determined by health condition scales) have a greater amount of contact with their family than do those in poorer health. These findings are corroborated by others (Preston, 1986; Yang & Zhang, 1992; Field, Minkler, Falk, & Leino, 1993; Lin, 1992; Tu, Liang, & Li, 1989; Sugisawa et al., 1994; Colantonio et al., 1993; Wolinsky et al., 1993; Garrard, Buchanan, Ratner, Makris, Chan, et al., 1993; Engle and Graney, 1993; Feldblum, 1985).

Variables related to individual factors are clearly the best predictor of subjects' biopsychosocial functioning in this study. Individual predictors include current health conditions, gender, number of sources of insurance, reasons for institutionalization, native place, and age. Subjects with a greater number of insurance sources had better health outcomes; however, they had poorer health conditions on admission. Yang's (1992) study of the elderly with self-provided insurance in nursing homes in Taiwan showed that individuals with no health insurance (self-pay) had poor health conditions and higher health care needs, and thus were more likely to be admitted earlier. Individuals who have more insurance sources may have less financial anxiety which may affect their psychosocial health, and thus result in a higher score for biopsychosocial functioning. Accordingly, people with more insurance sources may be in the higher social class, a factor which has been found to be strongly and consistently associated with health care outcomes (Ory, Abeles, & Lipman, 1992). Since social class is positively related to a stable living environment, availability of coping options, and a network of people with resources similar to their own (Ory et al., 1992), these individuals are at less risk for institutionalization than individuals who have no insurance source.

Subjects with family members who were not willing to care for them had better functional status than those either with no family members or those with family members who were not able to take care of them. Subjects with no family members experience less social support, increased psychological distress, greater loneliness, and probably experience worse biopsychosocial functioning than other groups. Subjects whose family members are not willing to take care of them may lack positive family relationships between family members. This population tends to have a higher risk of institutionalization regardless of their functional status. Feldblum (1985) confirmed this point by finding that the major factor in keeping the elderly in the community is not the degree of functional impairment but the level of access to family care and social relationships. Individuals having poor family relationships may seek admission earlier even though there is no decline in functional status.

Satisfaction with care was the best predictor of the outcome variable, biopsychosocial functioning. When individual and organizational factors were controlled, both satisfaction measures significantly mediated the effect of individual and organizational factors on subjects' biopsychosocial functioning, yet no relationship between these two scores was detected. One explanation for the lack of correlation between the satisfaction measures may be due to the interaction of culture on the data collection. By tradition Chinese are conservative and would be unwilling to be critical of the staff but would be more likely to criticize the global operation of the institution. Furthermore, since the subjects had no prior knowledge of the investigator, they may have questioned her true intent in gathering the required information. In general, the self-reported levels of satisfaction can also be considered an outcome indicator of the quality of nursing home care. As expected, the more physicians were involved in the subject's care, the higher the patient's score on the multidimensional functioning. Subjects' physical and social, and psychological and social functioning were positively and significantly related in this study, whereas contrary to expectation, no relationship between physical and psychological functioning was detected.

This study addressed and measured specific nursing-sensitive patient outcomes (biopsychosocial functioning and satisfaction with care) in a nursing home population in Taiwan. The positive changes in subjects' health conditions after admission to a nursing home is cited by many authors (Stein, Linn, & Stein, 1985; Chenitz, 1983; Mohlenkamp, Gress, & Flood, 1975), further documenting that health status deterioration does not always occur during a stay in these settings. All health condition variables showed improvement in this research. Improvement occurred in activities of daily living, mental and behavior conditions, and decreased need for medical interventions. The increase of rehabilitative interventions after admission may imply that adequate rehabilitation facilities were available in the study settings. This also indicates that the nursing staff may have recognized patients' rehabilitation needs and reinforced those activities which prevent a decline in physical ability or fostered those activities which result in greater independence and an improvement in activities of daily living after admission.

The documentation of changes in health conditions in the study population provides general recognition for nursing practice regardless of practice pattern. Eating, transferring, dressing, toileting, rehabilitative interventions, mental and behavior conditions are greatly impacted by nursing interventions. In long-term care facilities, outcomes of patient care are more likely to be sensitive to the effects of nursing interventions. Nursing homes are relatively untouched by medical practice, even though the number of physicians involved did contribute to patient satisfaction in this study. Nurses will continue to be in demand to provide therapies that physicians cannot or will not do. From a policy perspective, this is the perfect time for professional nursing to develop in nurse managed institutions.

Authors

Ke-Ping A. Yang, PhD, RN
E-mail: agnes@vghtc.vghtc.gov.tw

Ke-Ping A. Yang, PhD, RN is a nursing supervisor at Taichung Veterans General Hospital and an Associate Professor at Chung-Shan Medical College in Taichung, Taiwan.

Lillian M. Simms, PhD, RN, FAAN
E-mail: lsimms@umich.edu

Lillian M. Simms, PhD, RN, FAAN is an Associate Professor of Nursing Emeritus of The University of Michigan, School of Nursing, Ann Arbor, Michigan, USA.

Jeo-Chen T. Yin, PhD, RN
E-mail: jcyin@vghtpe.gov.tw

Jeo-Chen T. Yin, PhD, RN is Director of Nursing Department, Taipei Veterans General Hospital, and an Associate Professor, at the National Yang-Ming University in Taipei, Taiwan.