Sociodemographic variables predicting poor post-discharge outcomes for hospitalized elders with heart failure

Sociodemographic variables predicting poor post-discharge outcomes for hospitalized elders with heart failure

Paula Roe-Prior

Heart failure, a clinical syndrome characterized by progressive cardiac dysfunction, affects 10 in 1,000 Americans over age 65 (American Heart Association, 2006). In 2006, the estimated direct costs of heart failure in health care expenditures was $26.8 billion, with most of that cost attributable to hospitalizations and some of it possibly preventable (Osterberg & Blaschke, 2005). Type, quantity, and quality of health service use vary by certain sociodemographic characteristics, such as age, race, education, gender, income, marital status, and living situation (Arozullah et al., 2006; Asch et al., 2006; Davis, Evans, Strickland, Shaw, & Wagner, 2001; Gan et al., 2000; Hwang, Ryan, & Zerwic, 2006; Iwashyna & Christakis, 2003; Rosenfeld, Lindauer, & Darney, 2005; Shah, Rathouz, & Chin, 2001; Sun, Burstin, & Brennan, 2003; Zerwic, Ryan, DeVon, & Drell, 2003). However, with the exception of age and possibly living situation, discharge planning seldom incorporates sociodemographics to identify hospitalized heart failure patients in need of post-discharge services.

Recent research has demonstrated that sociodemographics are more important for predicting quality and type rather than quantity of health care received (Asch et al., 2006; Kossovsky et al., 2004). The study by Asch and colleagues (2006), which reviewed the medical records of a random sample who had at least one visit with a health care provider, found little difference based on sociodemographics in quality of care. All groups received less than ideal care, but more highly educated, wealthier, younger, female patients were more likely to have received better quality of health care. Another study (Kossovsky et al., 2004) assessed the inpatient quality of care of 371 heart failure patients over age 65. Researchers reported the oldest old were the least likely to receive appropriate evaluation and treatment. Because nurses have primary responsibility for performing discharge planning, knowledge of the sociodemographics known to influence service use is essential.

Use of sociodemographics known to predict service use with more accuracy may allow patients to be directed to the appropriate level of health care and prevent the more costly use of acute care services, such as hospitalizations, emergency department (ED) visits, and unscheduled physician office and clinic visits. The purpose of this study was to investigate sociodemographic factors as predictors of poor post-discharge outcomes in hospitalized elders. In this study, only variables previously identified in the literature for service use prediction were examined. These variables included age, gender, race, living situation, education, marital status, and income. Poor post-discharge outcomes were defined as rehospitalizations, ED use, and unscheduled acute care physician office or clinic visits.


The purpose of this study was to perform a secondary analysis of data collected in an earlier study (Roe-Prior, 2004) to determine if predisposing factors, such as age, gender, race, living situation (alone or with family or friends), marital status, education, and income were related to post-discharge service use (rehospitalizations, ED visits, and acute unscheduled physician office or clinic visits) for elders hospitalized with an acute exacerbation of heart failure.


Data collection. This study pooled data from the control group heart failure patients from The Comprehensive Discharge Planning Studies for Hospitalized Elders (Naylor et al., 1994; Naylor et al., 1999) admitted to two Philadelphia (urban) hospitals and a similar group of elders hospitalized with heart failure admitted to two Scranton, Pennsylvania, (community) hospitals (Roe-Prior, 2004). For all patients, during index hospitalization, demographic and clinical data were collected via an in-person interview. A committee of heart failure experts reviewed the data collection tool, and the ease of use of the tool was established in a pilot study. Patients were contacted by phone at 2, 6, and 12 weeks after discharge and data were obtained on the use of acute health services during those time periods.

Sample. Informed consent to review patients’ records was obtained previously for the original study, and the study was approved by the appropriate institutional review boards. Eligibility criteria included age 65 and older; alert and oriented at admission; admitted from home (because these subjects were more likely to be discharged to home); English speaking and able to respond to questions; able to be reached by telephone after discharge; and discharge medical diagnosis of congestive heart failure. Forty-eight of the patients were enrolled at two Philadelphia hospitals and 55 at two Scranton hospitals, yielding a final sample size of 103.

Analysis. All analyses were performed using the SPSS[R] (2000) statistical software. Service users versus non-service users were compared by calculating frequencies and performing chi square tests on nominal level data and Student’s t tests on interval level data. Multiple bivariate correlations were performed to identify the interrelatedness of the dependent variables (DVs), the independent variables (IVs) that were multicollinear, and the IVs that had a significant correlation with the DVs. This was done to construct the regression equations with the most parsimonious and least redundant IVs, sociodemographic factors, and acute post-discharge service use as well as to identify the independent variables that were multicolinear.

Separate multiple regression equations were calculated for each type of service used over the entire 12-week period: all-cause rehospitalization, heart failure-related readmissions, unscheduled physician office or clinic visits, ED visits, and total services used. Backward stepwise regression was used to eliminate the weakest predictors of each type of service use, with alpha to remove at 0.10 (Munro & Page, 1993; Tabachnick & Fidell, 1996).


Participants were primarily female (55.3%), white (69.9%), unmarried (55.4%), and not working (92.2%). The average age was 78, with an average length of index hospital stay of 7 days. The majority of subjects had a high school diploma or less (71.6%) and a yearly income of less than $20,000 (67.9%) (see Table 1). In contrast to urban subjects, subjects from the community hospitals were all white and primarily male. No statistically significant differences existed on the other predisposing factors of age, marital status, education, employment, or income in patients between the two sites (see Table 2). Difference in length of stay was not significant between the two sites. Subjects excluded from the analysis due to missing data had more impaired functional status (p=0.023) and more coexisting conditions (p=0.042) than those who completed the study.

Over the 12-week period, 43 patients had all-cause rehospitalizations (total of 57 readmissions). Multiple regression was computed for all-cause rehospitalization. In the model with the best fit, being unmarried predicted all-cause rehospitalization, while being of low income approached significance (p=0.056) (see Table 3). Of the 43 patients rehospitalized over the 12-week period, data were available on the cause of rehospitalization for 42 patients. Of these, 34 readmissions were related to heart failure. The only variable correlated significantly with readmissions related to heart failure was having an ED visit (r=0.77, p<0.001), although only ED visits that did not result in hospitalization were counted for the purpose of this analysis.

Thirty-seven patients had acute unscheduled physician visits over the 12-week period. In the univariate analysis, only admission to an urban hospital (r = -0.35, p=<0.001) and black race (r=0.35, p<0.001) significantly correlated with the need for a physician visit. None were significant in the multivariate analysis. Over the 12-week period, 18 patients had a total of 22 ED visits. In the univariate analysis, the IV significant for predicting ED visits was community site (r=0.19, p=0.044). Both lower education and community site were significant in the multivariate analysis (see Table 3). A total of 151 services, including all-cause rehospitalizations, physician visits, and ED visits, were used by 65 patients. In the multivariate analysis, black/Asian race, and lower income were correlated positively with using more services.


Results of this study confirm previous research findings: sociodemographic factors, while less important than severity of illness factors, play a role in predicting post-discharge service use. All-cause rehospitalization was predicted by being unmarried, while low income approached significance as a predictor. Others have reported that married elders and those of higher income are more likely to have a primary care provider and use preventive health services (Culica, Rohrer, Ward, Hilsenrath, & Pomrehn, 2002; Gornick, Eggers, & Riley, 2004; Morales et al., 2004). Patients with more exposure to health care providers after discharge may be more adept at managing their heart failure and receive treatment for coexisting conditions, preventing hospital readmission. Because the study was done prior to initiation of the Medicare prescription drug program, low-income individuals ineligible for other drug assistance programs may have been unable to afford the costs of medication, leading to an exacerbation of their disease. Low income was a significant predictor of using more total services. In addition, married elders may benefit from their spouses’ reminders to follow their therapeutic regimen and thus avoid rehospitalization.

Although none of the factors were predictive of heart failure-related readmissions, severity of illness indicators for this population may be more relevant than sociodemographics at predicting hospitalization. Braunstein and colleagues (2003) found that co-morbidities were common in patients with heart failure, with 39% of their sample suffering from five or more co-morbidities while only 4% had uncomplicated heart failure. Also the current study did not evaluate how many of the physician and ED visits were heart failure-related and possibly prevented more costly hospital admissions.

Black/Asian race was correlated significantly to the use of unscheduled physician visits. Race may not affect access to health care (Chen, Rathore, Radford, Wang, & Krumholz, 2001; Skinner, Weinstein, Sporer, & Wennberg, 2003). As a result blacks may receive neither the routine follow up needed for disease self-management nor the appropriate treatments. This is further illustrated by the earlier finding (Roe-Prior, 2004) that fewer urban patients, who were primarily black, were likely to be discharged on angiotensin-converting enzyme (ACE) inhibitors than the white patients admitted to the community hospitals.

Community dwellers were more likely to use the ED for acute care than urban patients, and patients who used the ED for acute care tended to be less educated. It may be that less-educated patients allow their heart failure symptoms to become too severe for management in an office or clinic yet do not require hospitalization. Although not specific to patients with heart failure, a study of ED use (Sun et al., 2003) found that having a high school education or less, or being unmarried or a single parent were predictors of frequent ED use. A similar study of noninstitutionalized elders reported the variables predicting ED use were older age, lower education, and living alone (Shah et al., 2001). An explanation for the relationship of community dwelling and ED use may be that these patients may not incur the long waits which would dissuade urban dwellers from using an ED over a clinic or physician office.

Total services use was predicted by low income and race. These findings are in agreement with other authors (Gilligan, Kneusel, Hoffman, Greer, & Nattinger, 2002; Gornick et al., 1996) who reported that while persons of low income and blacks were less likely to use preventive services, they were at higher risk for complications from their diseases and would require the use of more acute care services.


This study may have been powered insufficiently to detect the effects of predisposing factors on resource use since power and effect size analyses were calculated for the original study. The original study was powered at 90% at a moderate effect size. Also, other variables such as attitudes toward health and health care providers were not measured. Replicating the study in a larger sample and measuring severity of illness, sociodemographics, and quality of care may provide a better predictive model of elders at risk to use acute post-discharge health services.

The amount of home care used by the sample also was not measured. Receiving home care may have affected the use of acute care services by this group positively or negatively. Because the study was done prior to regulatory requirements that all heart failure patients receive smoking cessation counseling, ACE inhibitors, and ejection fraction measurement, the current influence of sociodemographics may be somewhat attenuated.


The complexity of predicting acute post-discharge service use for elders hospitalized with heart failure is evident from the results of this study. Another study testing the predictive value of severity of illness, sociodemographics, and patient preference may provide a more complete profile of elders at high risk for acute post-discharge service use. It may then be possible to develop a scoring tool to stratify patients’ risks for poor post-discharge outcomes. Based on this tool, interventions could be designed to target patients to the appropriate level of follow-up care. More comprehensive services may be required for some elders based on their profiles, while others may be identified who require less-intensive services, and still others may require palliative care. However, any new intervention must be evaluated for cost effectiveness.

Further research is needed to risk stratify elders with heart failure and direct them to the most appropriate level of health care service with the potential to improve patient outcomes while conserving health care resources. Combining an analysis of easily obtained sociodemographic variables is useful for identifying patients at risk for poor post-discharge outcomes. Because nurses often perform discharge planning, they need to be aware that severity of illness alone may provide an incomplete picture of an elder’s risk for poor post-discharge outcomes. Sociodemographics also play a role in an elder’s ability to manage this complex syndrome. *


American Heart Association. (2006). Economic cost of cardiovascular diseases. In Heart disease and stroke statistics–2006 update. Retrieved August 15, 2007, from http://www.american 64858055-1026_HS_Stats06book.pdf

Arozullah, A.M., Lee, S.D., Khan, T., Kurup, S., Ryan, J., Banner, M., et al. (2006). The roles of low literacy and social support in predicting the preventability of hospital admission. Journal of General Internal Medicine, 21, 10-145.

Asch, S.M., Kerr, E.A., Keesey, J., Adams, J.L., Setodji, C.M., Malik, S., et al. (2006). Who is at greatest risk for receiving poor-quality health care? The New England Journal of Medicine, 354(11), 1147-1156.

Braunstein, J.B., Anderson, G.E, Gerstenblith, G., Weller, W., Niefeld, M., Herbert, R., et al. (2003). Noncardiac comorbidity increases preventable hospitalizations and mortality among Medicare beneficiaries with chronic heart failure. Journal of the American College of Cardiology, 42(7), 1226-1233.

Chen, J., Rathore, S.S., Radford, M.J., Wang, Y., & Krumholz, H.M. (2001). Racial differences in the use of cardiac catheterization after acute myocardial infarction. The New England Journal of Medicine, 344(19), 1443-1449.

Culica, D., Rohrer, J., Ward, M., Hilsenrath, P., & Pomrehn, P. (2002). Medical checkups: Who does not get them? American Journal of Public Health, 92(1), 88-92.

Davis, L.L., Evans, J.J., Strickland, J.D., Shaw, L.K., & Wagner, G.S. (2001). Delays in thrombolytic therapy for acute myocardial infarction: Association with mode of transportation to the hospital, age, sex, and race. American Journal of Critical Care, 10(1), 35-42.

Gan, S.C., Beaver, S.K., Houck, P.M., MacLehose, R.E, Lawson, H.W., & Chan, L. (2000). Treatment of acute myocardial infarction and 30-day mortality among women and men. The New England Journal of Medicine, 343(1), 8-15.

Gilligan, M.A., Kneusel, R.T., Hoffman, R.G., Greer, A.L., & Nattinger, A.B. (2002). Persistent differences in sociodemographic determinants of breast conserving treatment despite overall increased adoption. Medical Care, 40(3), 181-189.

Gornick, M.E., Eggers, P.W., Reilly, T.W., Mentnech, R.M., Fitterman, L.K., Kucken, L.E., et al. (1996). Effects of race and income on mortality and use of services among Medicare beneficiaries. The New England Journal of Medicine, 335(11 ), 791-799.

Gornick, M.E., Eggers, P.W., & Riley, G.F. (2004). Associations of race, education, and patterns of preventative service use with stage of cancer at time of diagnosis. Health Services Research, 39(5), 14031427.

Hwang, S.Y., Ryan, C., & Zerwic, J.J. (2006). The influence of age on acute myocardial infarction symptoms and patient delay in seeking treatment. Progress in Cardiovascular Nursing, 21(1), 20-27.

Iwashyna, T.J.,& Christakis, N.A. (2003). Marriage, widowhood, and health-care use. Social Science and Medicine, 57, 2137-2147.

Kossovsky, M.P., Sarasin, F.P., Louis-Simonet, M., Chopard, P., Sigauld, P., Perneger, T.V., et al. (2004). Age and quality of in-hospital care of patients with heart failure. Europena Journal of Public Health, 14(2), 123-127.

Morales, L.S., Rogowski, J., Freedman, V.A., Wickstrom, S.L., Adams, J.L., & Escarce, J.J. (2004). Use of preventative services by men enrolled in Medicare + Choice Plans. American Journal of Public Health, 94(5), 796-803.

Munro, B.H., & Page, E.B. (1993). Statistical methods for health care research (2nd ed.). Philadelphia: J.B. Lippincott.

Naylor, M.D., Brooten, D., Campbell, R., Jacobsen, B.S., Mezey, M.D., Paulay, M.V., et al. (1999). Comprehensive discharge planning and home follow-up of hospitalized elders. JAMA, 281(7), 613620.

Naylor, M.D., Brooten, D., Jones, R., Lavizzo-Mourey, R., Mezey, M., & Pauley, M.(1994). Comprehensive discharge planning for the hospitalized elderly. Annals of Internal Medicine, 120(12), 999-1006.

Osterberg, L., & Blaschke, T. (2005). Drug therapy: Adherence to medication. The New England Journal of Medicine, 353(5), 487-497.

Roe-Prior, P. (2004). Variables predictive of poor post-discharge outcomes for hospitalized elders in heart failure. Western Journal of Nursing Research, 26(5), 533-546.

Rosenfeld, A.G., Lindauer, A., & Darney, B.G. (2005). Understanding treatment-seeking delay in women with acute myocardial infarction: Descriptions of decisionmaking patterns. American Journal of Critical Care, 14(4), 285-293.

Shah, M.M., Rathouz, P.J., & Chin, MH. (2001). Emergency department utilization by noninstitutionalized elders. Academic Emergency Medicine, 8(3), 267-273.

Skinner, J., Weinstein, J.N., Sporer, S.M., & Wennberg, J.E. (2003). Racial, ethnic, and geographic disparities in rates of knee arthroplasty among Medicare patients. The New England Journal of Medicine, 349(14), 1350-1359.

Sun, B.C., Burstin, H.R., & Brennan, T.A. (2003). Predictors and outcomes of frequent emergency department users. Academic Emergency Medicine, 10(4), 320-328.

Tabachnick, B.G., & Fidell, L.S. (1996). Using multivariate statistics (3rd ed.). New York: Harper Collins College.

Zerwic, J.J., Ryan, C.J., DeVon, H.A., & Drell, M.J. (2003). Treatment seeking for acute myocardial infarction symptoms. Differences in delay across sex and race. Nursing Research, 52(3), 159-167.

Paula Roe-Prior, PhD, RN, is an Assistant Professor, University of Scranton, Department of Nursing, Scranton, PA.

Table 1.

Group Demographics

N Mean Percent

Age 103 77.65


White 72 69.9

Black/Asian 31 30.1


Male 46 44.7

Female 57 55.3

Marital Status

Married 44 44.6

Widowed/Single 59 55.4


Less than high school 42 41.2

High school diploma 31 30.4

Post high school 29 28.4


Working 8 7.8

Not working 95 92.2


< $10,000 40 38.8

$10,001-$19,999 30 29.1

> $20,000 23 22.3

Don’t know/No Answer 10 9.7

Table 2.

Comparison of Urban and Community Subjects on

Demographic Variables

Urban Community

N Percent Percent p Value

Race 103

White 35.4 100 0.0001 *

Black/Asian 64.6 0

Sex 103

Male 31.2 56.4 0.011 *

Female 68.8 43.6

Marital Status 103

Married 33.3 50.9 0.072

Widowed/Single 66.7 49.1

Education 102

Less than high school 46.8 36.4 0.505

High school diploma 25.5 34.5

Post high school 27.7 29.1

Employment 103

Working 8.3 7.3 0.841

Not working 91.7 92.7

Income 103

Less than $10,000 45.8 32.7 0.483

$10,001-$19,999 22.9 34.5

$20,000 or more 20.8 23.6

Don’t know/No Answer 10.4 9.1

Age 103

Mean 78.04 77.31 0.520

Standard Deviation 5.97 5.50

Length of Stay 103

Mean 7.50 6.69 0.420

Standard Deviation 6.19 3.25

* p < 0.05

Table 3.

Multiple Regression Results for Predictors of Service Use

R2 R2 Change df F

Rehospitalization 0.127 0.085 (5,90) 2.478

Marital status (widowed/


Low income

Site (community)



ED Visits 0.079 0.079 (2,99) 4.272


Education (less than

high school)

Total Services 0.098 0.098 (3,99) 3.589



Income (lower)

[beta] p 95% CI

Rehospitalization 0.252 0.021 0.060 -0.716

Marital status (widowed/

unmarried) 0.172 0.059 -0.007 -0.351

Low income 0.160 0.122 -0.029 -0.240

Site (community) -0.165 0.121 -0.050 -0.006

Age -0.182 0.109 -0.373 -0.038


ED Visits 0.210 0.032 0.019 -0.417

Community -0.240 0.038 -0.248 -0.007

Education (less than

high school)

Total Services 0.211 0.030 0.023 -0.422

Race -0.148 0.124 -0.609 -0.075

Employment 0.192 0.047 0.001 -0.184

Income (lower)

COPYRIGHT 2007 Jannetti Publications, Inc.

COPYRIGHT 2008 Gale, Cengage Learning