Heart disease and pulmonary disorders are among the primary reasons for hospitalization and home care referrals among hospitalized Medicare beneficiaries (Helberg, 1993; MorrowHowell & Proctor, 1994). As a result of shorter lengths of hospital stay, the burden and cost of caring for older adults with chronic obstructive pulmonary disease (COPD) and congestive heart failure (CHF) have shifted from hospitals to patients' homes. Currently, Medicare guidelines stipulate that home care following hospital discharge is reimbursed only when patients are homebound and demonstrate the need for skilled nursing and physical, occupational, or speech therapy (Health Care Financing Administration, 1994). These guidelines have been described as potential "barriers to care [that] cause underuse of services" (Pohl, Collins & Given, 1995, p. 34).
Failure to make timely and appropriate home care referrals has been linked to poor post-discharge outcomes, including preventable hospital readmissions. Readmissions are a particularly challenging problem for patients discharged with COPD or CHF (American Thoracic Society [ATS] Statement, 1995; Rich et al., 1993; Venner & Seelbinder, 1996).
To date, research in this area has focused on improving the quality of discharge planning and promoting positive post-discharge outcomes for hospitalized elderly individuals (Naylor et al., 1994). For example, in a recent study, an advanced practice nurse (APN) directed discharge planning and home care intervention for hospitalized older adults at risk for rehospitalization. These actions reduced readmissions, lengthened the time between discharge and readmission, and decreased the costs of providing health care (Naylor et al., 1999). In a group of patients who had been hospitalized for exacerbation of COPD, a majority of patients (81.4%) were rehospitalized following the index hospitalization (i.e., hospitalization when patients were enrolled in the studies), primarily for respiratory and cardiac symptoms (Narsavage, Grant, & Dolan, 1998). However, there was a significantly (p < 0.05) longer length of stay (LOS) at home before rehospitalization for participants who received home care nursing than for those who did not. While multiple factors, including positive outcomes, could justify the need for home follow-up after hospital discharge and while the decision to refer for home care is an integral part of the discharge planning process, little is known regarding factors associated with referrals for home care.
Because of the major changes occurring in health care, the authors have limited the literature review to the more recent studies on discharge planning. Exploratory research suggests that clinicians can successfully identify those who do not need home care [non-need], but that there remains a group of patients in need of home care who are not identified [non-referred] (Prescott, Soeken, & Griggs, 1995). In a convenience sample of 145 participants hospitalized at least 48 hours for a medical or surgical condition, 36 were referred for home care. Home care referrals were more common in patients who were older and had longer lengths of hospital stays, greater number of diagnoses, previous hospital admissions, and higher levels of dependency with lower levels of physical functioning, social support, and readiness for selfcare. At higher than average risk for poor outcomes were those nonreferred in need of home care - older patients who had lower intent to adhere, less social support, and a longer hospital LOS - almost 26% of the sample.
A few studies have examined patient characteristics that may contribute to the need for home care after hospital discharge. Cloonan and Belyea (1993) examined 352 records of patients referred for skilled nursing after hospital discharge. In this study, functional limitations, cognitive impairment, medical diagnoses, age, and LOS were found to be related to the type and amount of home care received. Number of comorbid conditions (Solomon et al., 1993), previous hospital admissions (Prescott et al., 1995), complex medication regimens (Weaver & Burdi, 1992), use of oxygen (Narsavage, 1996; Schultz, Stark, & Petro, 1996), as well as functional deficits (Narsavage, 1996) have also been associated with the need for home care after discharge.
Other studies have demonstrated that multiple factors in addition to patient characteristics relate to the type and amount of home care needed. Despite similarities in physiological status and identical diagnoses, some patients with chronic illness consistently function better than others and need less assistance (Narsavage & Weaver, 1994; Weaver & Burdi, 1992). Research findings suggest that characteristics such as the living situation or marital status (Granger, Divan, & Fiedler, 1995; Narsavage, 1996; Pohl et al., 1995) and payment source (Experton, Li, Branch, Ozminkowski & MellonLacey, 1997) may also contribute to the need for home care referral.
The identification of an enhanced set of factors that can be used to make appropriate and timely home care referrals for older adults hospitalized with the cardiac or pulmonary disorders could contribute to improved post-discharge outcomes for these vulnerable patient groups. There is overlap in these disease conditions such that dual diagnoses of both CHF and COPD may be present. Clinicians have suggested that patients with dual diagnoses are at greater risk. No home care referral profile has been developed for this elderly patient population. By performing a secondary analysis of data from patients with an admission diagnosis of CHF or COPD, those with a dual diagnosis could be compared with those having a single disease. The purpose of this study was to examine patient factors, including single or dual diagnosis, associated with health care professionals' referrals of elderly individuals for home care immediately following hospitalization for CHF or COPD.
This study was a secondary analysis of patient records and records of interviews of 159 adults age 65 years and older. Data were drawn from three studies. Two of the studies included older adults hospitalized with a primary diagnosis of CHF (diagnosis related group [DRG] 127) and adults hospitalized with an "atrisk condition" (e.g., CHF) plus at least one other risk criteria (e.g., older than age 80, low social support, multiple comorbidity, functional impairment, poor self-rating of health, history of depression, previous hospitalizations, or nonadherence). The third study was comprised of post-hospitalized patients with COPD (DRG 88). The first study examined the effects of a discharge planning protocol. This protocol was implemented by APNs for older adults hospitalized with selected cardiac conditions (Naylor et al., 1994). The second study examined the effectiveness of a comprehensive discharge planning and home follow-up protocol implemented by APNs and designed for high-risk elderly individuals hospitalized with conditions that were among the top reasons for hospitalizations and home care use by Medicare beneficiaries in 1992 (e.g., CHF, angina, myocardial infarction, respiratory infections, coronary bypass surgery, cardiac valve replacement, bowel problems, and lower extremity orthopedic procedures) (Naylor et al., 1999). Because these two discharge planning studies were randomized clinical trials, only data obtained from control patients were included in this analysis.
In both discharge planning studies, participants were older adults, primarily women (67%), Black (64%), and cared for in large urban teaching hospitals. The third study examined post-hospitalization status following hospital admission for an exacerbation of COPD (Narsavage, 1996). Older adults in this study were men (52%), White, and primarily cared for in smaller community hospitals. Human subject approval had been obtained in each of the original studies.
The prevalence of study participants with both CHF and COPD diagnoses in the original three study samples resulted in reclassifying the combined sample into three diagnostic groups. The secondary analysis thus examined patients with CHF (N = 46, 74% women), patients with COPD (N = 45, 56% women), and patients with both COPD and CHF (N = 68, 44% women, 18% Black).
Based on the review of relevant research, variables that may be predictive of the need for home care following discharge were identified. These included: primary diagnosis, number of comorbid conditions, age, gender, ethnicity, marital status, insurance coverage, hospitalization within previous 6 months, length of index hospital stay, number of discharge medications, need for home oxygen and physical therapy (PT), and need for assistance with activities of daily living (ADLs). Data on these variables were originally recorded from patients' medical records and patient interviews as noted. Secondary files were constructed from merged Statistical Package for the Social Science (SPSS) files of the data originally recorded on standardized patient interview forms and hospital records. Range checks and checks for internal consistency were performed on the merged data set. For example, reclassifying the combined sample into three diagnostic groups controlled for source and ethnicity. The between group variation in age range for CHF patients was 65 to 89 years; for CHF patients, the range was 65 to 88 years; and for patients with both COPD and CHF, age ranged from 65 to 90 years. Data on number of comorbid conditions, number of discharge medications, need for home oxygen, PT, and assistance with ADLs were obtained through review of the original data collection forms. Internal consistency testing measured correlations between groups and among groups. For example, Medicaid status and need for a physical therapist were eliminated in this analysis as they were highly correlated and redundant with diagnosis and the need for a home health aide.
DEMOGRAPHIC CHARACTERISTICS OF PATIENTS WITH COPD1 CHF, AND BOTH COPD AND CHF
Measurement of Variables
Comorbid conditions were listed as the total number of nine possible health conditions in addition to COPD and CHF (i.e., diabetes, hypertension, arthritis, eye, heart, peripheral vascular, gastrointestinal, renal, other disorders) identified by patients at the index hospitalization as ongoing problems. Medications were originally listed as the total number of drugs prescribed at the index hospital discharge. These were subsequently recoded into commonly prescribed categories and dichotomized, based on the median number of prescribed drugs (N = 7). Length of stay for the index hospitalization was originally listed as number of days but was recoded for this study as a categorical variable (1 = less than 6 days, 2 = 6 or more days). Less than 6 days was selected because it was the expected arithmetic mean hospital LOS for these DRGs (COPD = 5.7, CHF = 5.8 days). Marital status (yes or no) served as a measure of social support (Pohl et al., 1995). Gender, race (White/Black), Medicaid status, diagnosis (COPD, CHF, or both), need for oxygen, need for home health aide to assist with ADLs, need for physical therapist, and hospitalization within 6 months prior to index hospitalization were all dichotomous variables. Age was recorded as a quantitative variable. The dependent variable, home care referral prior to hospital discharge, was a dichotomous variable.
SUMMARY OF LOGISTIC REGRESSION ANALYSIS TESTING PREDICTORS OF RECEIVING HOME CARE NURSING POSTHOSPITAL IZATION
A binary dummy variable identifying the medical condition as a combined diagnosis of COPD and CHF or a single diagnosis was used to control their significant intercorrelations (r > 0.5, p < .0001) in the combined data set. The primary analytic method was a logistic regression to predict the odds of being referred for home care or not. There were 13 potential independent variables. The 13 independent variables were examined for their unique bivariate relationship to the dependent variable and to every other independent variable. Medicaid status and need for a physical therapist were eliminated in this analysis because of redundancy with diagnosis and need for home health aide. Dual and single diagnoses were entered first as the best initial predictive subset based on their high intercorrelation. In the second step, those variables that were independently and significantly correlated with the dependent variable were then entered simultaneously into the regression equation. The importance of each variable in the model was verified through the likelihood ratio test.
The mean age of the total sample of 159 participants was 76; 56% were women; 24% were Black; 37% were married; and 32% were receiving Medicaid. At the index hospitalization, 74% had been hospitalized at least once within the previous 6 months; 35% used oxygen at home; and 26% needed a home health aide to assist with ADLs. At the index hospital discharge, participants in the combined sample had a mean of four comorbid conditions and seven prescribed medications. Ninety-eight (62%) were referred for home care while hospitalized and 61 (38%) did not receive a home care referral. Significantly more of the patients with a dual diagnosis than a single diagnosis received referral for home care (Kruskal-Wallis/> < 0.001).
Table 1 presents the relative risks associated with the need for home care for each of the independent variables. Referrals for home care after hospital discharge were more likely for patients who were not married (p = 0.04), were using more than seven prescribed medications (p = 0.04), had a hospital LOS greater than 6 days (p = 0.008), needed a home health aide (p = 0.0001), and needed home oxygen (p = 0.001). Hospitalization within the previous 6 months approached significance (p = 0.08). The following factors were not predictive of referrals for home care: age (p = 0.23), gender (p = 0.19), race (p = 0.13), and number of comorbid conditions (p = 0.62).
The results of the logistic regression conducted are presented in Table 2. The model chi-square was statistically significant (χp 2 = 48.02, df=7tp < .00001). First, a dual medical diagnosis (i.e., presence of both COPD and CHF) was predictive of referral (p = 0.02). The single diagnosis of either COPD or CHF was not predictive of referral (p = 0.55). In addition the need for a home health aide (p = 0.001) and a marital status of "not" married (p = 0.05) were predictive of referral. Having a hospital LOS greater than the arithmetic average DRG days (more than 6) approached significance (p = .09). Because LOS was categorized (splitting at fewer than 6) rather than continuous to better fit a logistic regression, its near significance may be related to the slight variation from the DRG arithmetic means of 5.7 to 5.8 days for COPD and CHF. It can be considered an integral part of the equation because dropping it from the equations negatively affects the likelihood of the model. The combination of dual diagnosis, not married, need for a home health aide, and having a longer than average LOS correctly predicted 84% of patients who received home care referrals and 68% who did not, for an overall correct prediction rate of 78%. Low deviance in the likelihood of fit indicates the model reasonably fits the case.
A single diagnosis of COPD or CHF may not have been statistically significant because the relative complexity of a dual diagnosis would dominate a single diagnosis. The need to use oxygen at home has not been a reason for referral under current Medicare guidelines. Most of the teaching related to the use of oxygen is expected to be completed prior to discharge with home follow-up not considered skilled care. Additionally the use of oxygen was more prevalent in those patients with COPD who developed heart failure and, thus, became insignificant when the effect of the combined diagnosis of COPD and CHF was entered into the model. Recoding medications into common categories to eliminate the source differences and provide a categorical variable for the logistic regression resulted in the loss of variance that could have accounted for its insigniiicance in the final equation.
It can be concluded that a model using individuals with the combined diagnoses of COPD and CHF, who are not married, and need assistance from home health aides may be helpful in predicting referral for home care immediately following hospital discharge for older adults hospitalized with COPD or CHF. Lengths of hospital stay greater than the DRG arithmetic mean LOS for the COPD and CHF approached significance (p = 0.09) in the model and could also be useful as an indicator of home care referral decisions. Of these four risk factors, needing a home health aide for assistance with ADLs and being unmarried are supported by previous studies (Prescott, Soeken, & Griggs, 1995; Solomon et al., 1993). Clinicians associate the availability of a spouse in the home with social support. In contrast, being unmarried or living alone is more likely to result in a referral for home care (Narsavage, 1996). A longer than average hospital LOS and the dual diagnosis of COPD and CHF arc likely to be indicative of a patient needing skilled care. Only a small percentage of these patients were also insured by Medicaid. Studies with larger sample sizes are needed to determine if dual eligible insurance status is influencing the decision to refer for home care.
Unlike previous studies, number of comorbid conditions was not found to be a significant predictor of home care referral. In this study the accuracy of documentation of comorbid conditions in patient records is questionable. The recording of coexisting health problems for patients from one hospitalization to another was found to be highly variable. Patient self-report of other health problems may be a more accurate method of obtaining these data, but the reliability of recall data is questionable. Alternatively, the use of functional status instead of comorbid conditions may be more appropriate in this patient population. Functional status has been shown to be a stronger predictor of home care need and referral than number of medical diagnoses for hospitalized elderly individuals (Solomon et al., 1993).
Solomon et al. (1993) also suggested the use of home care increases as the number of risk factors increases with 76% of those having four risk factors requiring home care. This finding is consistent with the results of this study. The predictive model generated in this study included four risk factors and was able to correctly classify 76 patients who received referrals. Only 15 patients who were predicted not to need home care were observed to have received it. However, there were 19 patients predicted to need home care based on this model who did not receive it. This finding raises important questions about the influence of other factors in clinicians' decision to refer for home care and the effects of these decisions on patient outcomes.
Important limitations of this study included:
* Differences in the study samples.
* Possible regional variations in discharge planning practices.
Research is needed to examine how this model might be influenced by factors such as race or regional variations in discharge planning practices. In addition, studies examining the relationship between this and other predictive models and patient outcomes are an important area for further investigation.
Nurses and other discharge planners may find the model useful in targeting those at high risk for needing home care and early interventions. Alternatively, resources may be conserved if those at low risk are identified early. Referrals for home care services initiated prior to hospital discharge may prevent or delay hospital readmission or nursing home placement, especially for elderly individuals with multiple chronic conditions. Subsequent research is required to determine whether there are other patient factors, clinician factors such as knowledge of community resources, system factors such as designated discharge planners, or regional factors that impact referrals. With further research, increased understanding of the factors that contribute to appropriate referrals for home care after hospital discharge could enhance clinicians' decision-making, promote positive outcomes, and decrease the costs of caring for these vulnerable patient populations.
- American Thoracic Society statement: Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease. (1995). American Journal of Respiratory and Critical Care Medicine, 752(5 Suppl.),S78-S121.
- Cloonan, P.A., & Belyea, MJ. (1993). Limits of using patient characteristics in predicting home health care coordination. Western Journal of Nursing Research, /5(6), 742-751.
- Experton, B., Li, Z., Branch, L.G., Ozminkowski, R.J., & Mellon-Lacey, D.M. (1997). The impact of payor/provider type on health care use and expenditures among the frail elderly. American Journal of Public Health, 87(2), 201-206.
- Granger, C.V., Divan, N., & Fiedler, R.C. (1995). Functional assessment scales. A study of persons after traumatic brain injury. American Journal of Physical Medicine and Rehabilitation, 74(2), 107-113.
- Health Care Financing Administration. (1994). The Medicare 1994 handbook (USDHHS Publication No. HCFA 10050). Baltimore MD: Author.
- Helberg, J. (1993). Patients' status at homecare discharge. Image, 25, 93-99.
- Morrow-Howell, N., & Proctor, E. (1994). Discharge destinations of Medicare patients receiving discharge planning: Who goes where? Medical Care, 32, 486-497.
- Narsavage, CL. (1996). Functional outcomes after homecare nursing Phase I. The European Respiratory Journal, 9(SuppL 23), 35S.
- Narsavage, G.L., Grant M., & Dolan, L. (1998). Comparison of hospital réadmissions of COPD patients who did and did not receive home care nursing. The European Respiratory Journal /2(S27), 25S.
- Narsavage, G.L., & Weaver, TE. (1994). Physiological status, coping, and hardiness as predictors of outcomes in chronic obstructive pulmonary disease. Nursing Research, 43(2), 90-94.
- Naylor, M., Brooten, D., Jones, R., LavizzoMourey, R-, Mezey, M., & Pauly, M. (1994). Comprehensive discharge planning for the hospitalized elderly. Annals of Internal Medicine , 120, 999-1006.
- Naylor, M.D., Brooten, D., Campbell, R., Jacobsen, B.S., Mezey, M.D., Pauly, M.V., & Schwartz, J.S. (1999). Comprehensive discharge planning and home follow-up of hospitalized elders: A randomized clinical trial. Journal of the American Medical Association, 281, 613-620.
- Pohl, J.M., Collins, C, & Given, CW. (1995). Beyond patient dependency: Family characteristics and access of elderly patients to homecare services following hospital discharge. Home Health Care Services Quarterly, 15(4), 33-47.
- Prescott, A., Soeken, K., & Griggs, M. (1995). Identification and referral of hospitalized patients in need of homecare. Research in Nursing & Health, 18(2), 85-95.
- Rich, M., Vinson, J., Sperry, J., Spinner, L., Chung, M., & Davila-Roman, V. (1993). Prevention of readmission in elderly patients with congestive heart failure: Results of a prospective, randomized pilot study. Journal of General Internal Medicine, 8, 585-590.
- Schultz, K., Stark, H., & Petro, W. (1996). Development of a standardized training program for patients with an indication for long-term oxygen therapy (LTOT). American Journal of Respiratory and Critical Care Mediane, 153(4), A783.
- Solomon, D.H., Wagner, D.R., Marenberg, M. E., Acampora, D., Cooney, L. M., & Inouye, S.K. (1993). Predictors of formal home health care use in elderly patients after hospitalization. Journal of the American Geriatria Society, 41, 961-966.
- Venner, G.H., & Seelbinder, J.S. (1996). Team management of congestive heart failure across the continuum./owm*/ of Cardiovascular Nursing, 10(2), 71-84.
- Weaver, F.M., & Burdi, M. (1992). Developing a model of discharge planning based on patient characteristics. Journal of Aging and Health, 4, 440-452.
DEMOGRAPHIC CHARACTERISTICS OF PATIENTS WITH COPD1 CHF, AND BOTH COPD AND CHF
SUMMARY OF LOGISTIC REGRESSION ANALYSIS TESTING PREDICTORS OF RECEIVING HOME CARE NURSING POSTHOSPITAL IZATION