Transitions in health status in older patients with heart failure.

Background: We aimed to determine transitions in health perception and functional status in older Medicare patients with heart failure.

Methods: We used 1991 to 1994 data from the Medicare Current Beneficiary Survey, a database that combines Medicare claims with yearly longitudinal surveys. We identified 872 patients 65 years or older in 1991 with a diagnostic code of heart failure.

Results: At baseline, 58% of the patients rated their general health perception as "fair" or "poor." Over 1 year, 18% of the patients died. Transition matrices revealed that health perception, activities of daily living, and instrumental activities of daily living were strong correlates of mortality; that dramatic changes in health status were relatively uncommon over 1 year among survivors; and that decline was common in patients with "excellent" or "very good" health perception. The prior year's health status and comorbidity were powerful predictors of the subsequent year's health status.

Conclusion: Many older patients with heart failure have worsening health status over time. Measures of prior health status can help predict chances of functional recovery.

Key Words: aging, functional status, health status, heart failure, quality of life

**********

Heart failure is a major public health problem that causes much morbidity and cost. (1) Many older patients have this condition, and heart failure is the most common cause for hospitalization of patients in the Medicare program in the United States. (2) Mortality and hospital readmission have been frequently studied outcomes for patients with heart failure. (3-5) However, especially for older patients who have heart failure and multiple medical problems, functional status and health-related quality of life (HRQOL) are particularly important endpoints.

Surprisingly little is known about how HRQOL evolves over time in older patients with heart failure. (6), (7) Most studies are either methodologic articles describing the development or validation of a scale, (8-16) or reports of randomized controlled trials and observational studies of drugs and nonpharmacologic interventions. (17-27) Although many studies have evaluated heart failure patients cross-sectionally, (28-32) few investigations have used health status measures longitudinally to follow the natural history of patients with heart failure. (33-36) Among those studies, Romm et al (33) studied 122 patients between the ages of 50 and 75 years for 6 months. They found that baseline health status, not process-of-care measures, were the most significant predictors of later symptom status. Diehr et al (37) noted sharp declines in health perception after an incident diagnosis of heart failure. Among survivors of an acute exacerbation of severe heart failure in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment (SUPPORT), health perceptions improved within 60 and 180 days. (35)

HRQOL information in a population-based heart failure cohort would be useful for public health planning purposes. Clinically, heart failure is often considered to be a high-morbidity condition, given its common symptoms of shortness of breath, fatigue, and weakness. However, national quantitative estimates of heart failure's burden on HRQOL are lacking, as is information on the likelihood of deterioration or recovery of HRQOL over time. National estimates of HRQOL would also provide both patients and physicians with a range of figures to consider as they discuss prognosis and what to expect as the patients live with their chronic disease. Clearly, such numbers would have significant limitations for individual clinical use, as prognostic estimates must be individualized and are difficult in any case with heart failure. (38) However, population-based data provide aggregate guideposts, and some of the clinical deterioration observed in community-based cohort data can be forestalled with the most optimal care. Therefore, in this study, we aimed to 1) determine changes in health perception and functional status in older patients with heart failure in a nationally representative cohort, 2) investigate whether baseline measures of HRQOL are associated with morbidity or mortality in patients with heart failure, and 3) identify factors associated with outcome.

Patients and Methods

Overview of the Data Source

The Medicare Current Beneficiary Survey (MCBS) is a database that combines Medicare claims data with longitudinal surveys of a nationally representative sample of approximately 14,000 beneficiaries in the Medicare program in the United States. (39) The Center for Medicare and Medicaid Services uses this database to assist with planning and management of the Medicare program. Whereas most Medicare database studies are limited by the relatively crude claims data, the MCBS supplements the national claims data with detailed survey information. The yearly survey of beneficiaries contains health perception, activities of daily living, instrumental activities of daily living, access to care, resource use, and comorbidities. The claims data include inpatient, outpatient, hospice, home health care, skilled nursing facilities, physician, and durable medical equipment. Thus, the MCBS is a powerful database that combines diverse information on a national sample of thousands of patients. The major drawback is that the level of clinical detail is limited to that found in the claims and patient survey. Thus, for example, detailed severity-of-illness data in medical records such as ejection fraction and coronary anatomy are not available. Nevertheless, for studies of HRQOL in a nationally representative sample of community-dwelling older adults with heart failure, we are unaware of a stronger data set. We used 1991 through 1994 data. Beginning in 1995, the MCBS replaced approximately 25% of its sample each year, complicating longitudinal studies. This project was exempt from human subjects regulations by the Institutional Review Board because it used existing data that are publicly available in which subjects cannot be identified.

Patient Population

Using the 1991 version of the MCBS Cost and Use file, we identified patients with an International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) diagnostic code of heart failure from the claims data: 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.93, 425.x, and 428 xx. (40) To increase the likelihood that a patient truly had heart failure, patients had to have either one inpatient code for heart failure or at least two outpatient codes for heart failure. Similar criteria with diabetes diagnostic codes have had an approximately 98% positive predictive value for identifying patients with diabetes. (41) We then followed this cohort for 3 years through the 1994 wave of the MCBS.

Survey Items

The baseline (1991) MCBS survey contains sociodemographic data including age, sex, race, and level of education. The key HRQOL outcome variables (health perception and functional status) were obtained in 1991 and also from the three follow-up waves of the MCBS survey (1992-1994).

Health Perception and Functional Status Outcome Measures. Different measures of health status capture different constructs of HRQOL. (42) In this study, we focus on three commonly used outcome measures of health and well-being, as follows.

1. General health perception: "In general, compared with other people your age, would you say that your health is (excellent = 1, very good = 2, good = 3, fair = 4, poor = 5)." (43)

2. The Katz Index of Activities of Daily Living (ADL) (44): "Because of a health or physical problem, do you have any difficulty bathing or showering, dressing, eating, getting in or out of bed or chairs, and using the toilet?" The response categories for these variables were "yes," "no," "doesn't do." "Yes" and "doesn't do" responses were collapsed, as "doesn't do" generally represents a disability for activities of daily living (ADLs). We summarized these data by computing and analyzing the total number (0-5) of activities for which the subject reported at least "any" difficulty.

3. Instrumental activities of daily living (IADLs) (45): "Because of a health or physical problem, do you have any difficulty using the telephone, doing light housework, doing heavy housework, preparing your own meals, shopping for personal items, managing money?" We collapsed "yes" responses and those who answered "doesn't do" because of a health or physical problem. Of these six activities, three (doing light housework, doing heavy housework, preparing your own meals) were deemed inapplicable for institutionalized patients by the designers of the MCBS, because these services are automatically provided for the patient. Thus, the MCBS does not ask about those three activities in institutionalized patients. To handle this, we do not analyze IADL responses for subjects living in institutions, and we complement the IADL model with a model for transition from a community into an institutional setting.

Medicare Claims Data

Complementing the survey information, the MCBS contains data from Medicare claims files. We created yearly comorbidity indicators by searching all claims' ICD diagnostic codes during the index year and each prior year in the data set. We analyzed the individual effects of specific diseases clinically pertinent to heart failure, including myocardial infarction (ICD 410.x), hypertension (ICD 401.x), and diabetes (ICD 250.x). We also obtained additional comorbid conditions from the claims data to construct an adaptation of the Deyo et al (46) and Romano et al (47) versions of the Charlson Comorbidity Index. The Charlson Comorbidity Index was originally validated to predict death in breast cancer patients, (48) and has been shown to be correlated with hospital readmission and death among patients with heart failure. (49) Heart failure and conditions studied individually (myocardial infarction, diabetes) were not included in the Charlson Comorbidity Index in this study to avoid double counting. The claims data also allowed us to construct annual variables for whether or not the subject resided in an institution at any time during the year.

Statistical Analysis

Development of Transition Matrices. To achieve specific aims 1 and 2, that is, determining changes in health perception and functional status and investigating whether baseline measures of HRQOL are associated with morbidity or mortality in older patients with heart failure, we created transition matrices that enable the reader to determine the chance a patient has of moving from one health state to another over time. The key outcomes of interest in this study are general health perception, ADLs, and IADLs in each of 3 years of follow-up. In addition, we also consider mortality and transition from living in the community to living in an institution. There are a variety of ways of summarizing and describing the trajectory over time of a cohort with respect to these outcomes. The approach we follow here is to capture these patterns using transition probabilities from one state to another over intervals of 1 year in duration, that is, from prior year t - 1 to year t, 1 year later, where t = 1992, 1993, and 1994. (50) For example, a subject reporting "good" general health perception in one year will have a certain probability of dying over the following year, a certain probability of surviving and reporting "excellent" health, remaining at "good" health, and so forth. Our statistical analysis focuses on modeling these transitions and presenting them in transition matrices expressing the probability of each possible state at year t for each possible state in the prior year t - 1. We present results stratified by age and sex.

We chose to model the 1-year health perception and functional status outcomes using ordinal logistic regression models (51) to calculate the probability of transitioning from one given health state to another, probabilities that make up the entries in the transition matrices. This approach avoids the problem of small cell sizes, and thus crude projections, that would arise if we simply took the observed proportions in each outcome as estimates of the outcome probability. In addition, we developed separate logistic regression models for the probability of transitioning to death. For example, for general health perception, we modeled mortality in year t as a function of general health perception and covariates in the prior year t - 1 in a logistic regression model. Then, conditional on survival to year t, we modeled general health perception in year t using an ordinal logistic regression model, again as a function of general health perception and covariates in the prior year t - 1.

Independent Associations with Health Perception and Health-Related Quality of Life. To achieve specific aim 3, identifying sociodemographic and clinical variables independently associated with health perception, ADLs, IADLs, institutionalization, and mortality, we used multivariate logistic and ordinal logistic regression models. (52) We used Stata, version 7.0 (Stata Corp., College Station, TX) for the analyses. (53) We present more details on the statistical analysis in the Technical Appendix.

Results

Patient Population

At baseline in 1991 adjusting for sampling weights, an estimated 62% of the patients were female, the mean age was 79 years, approximately 14% were nonwhite, and 54% had less than 12 years of education (Table 1). Approximately two-thirds had comorbidities and 58% of the patients rated their general health perception as "fair" or "poor." The average number (standard error) of deficiencies in ADLs and IADLs were 1.51 (0.06) and 2.15 (0.08), respectively, where the IADL figure refers to the noninstitutionalized sample at baseline. Of the 872 patients in the original cohort, 18% died within 1 year in weighted analyses. Table 2 shows the natural history of the cohort in unweighted analyses.

Transitions in Health Perception, ADLs, and IADLs

Table 3 shows estimated probabilities a patient has of transitioning from one health state to another over the course of 1 year for health perception, ADLs, and IADLs. For each health outcome, four matrices are presented stratifying by gender and whether the patient is 70 or 80 years old (all data used for estimates). The rows for each outcome variable represent the prior year's state, and the columns for each outcome variable represent the current state. The footnoted diagonal numbers (footnote e) in Table 3 represent no change in health status. Movement within a given t - 1 row below the diagonal represents an improvement in health status, and movement above the diagonal is a worsening in health status.

Several general points are apparent. First, worse health perception and more deficiencies in ADLs and IADLs are strongly correlated with mortality. For example, groups of 80-year-old women reporting excellent health perception have on average an 8% chance of transitioning to death the next year, whereas if they report good health perception they have a 13% chance of transitioning to death the next year, and if they note poor health perception they have a 25% chance of dying the next year. Second, large jumps in the outcomes are unusual among survivors. Most commonly, patients' health status remains unchanged or else may improve or worsen by one category. Third, sizable proportions of patients with excellent or very good health perception decline. For example, for 70-year-old men, 81% with excellent health perception and 62% with very good perception decline or die. In contrast, of 70-year-old men with zero ADL and zero IADL deficiencies, 25% and 37% decline or die, respectively.

Adjusting for the previous year's health status, age appears to have a few effects. General health perception is similar across age within gender strata, except for a higher rate of death among older people with poor or fair health perception. Across ADL strata, older women have a higher rate of mortality compared with younger women.

Patients Lost to Follow-up

We found that worse previous year's general health perception and worse previous year's IADLs predicted dropout in the following year, but previous year's ADL status was not predictive of dropout. To examine the sensitivity of the transition matrices to dropouts, and because it appeared as if subjects who are worse off are more likely to drop out, we performed simple sensitivity analyses by randomly assigning for each outcome half of the missing values to the highest (worst) level and half to the next highest level of that outcome. We reconstructed the transition matrices. We found the same general patterns as we reported above. For general health perception, all probabilities in the transition matrices changed by less than 5%. For ADLs, one probability changed by 11%, one by 8%, and one by 7%. For IADLs, one probability in the transition matrices changed by 10%, one by 8%, and four by 7%.

Independent Associations with Health Perception, ADLs, and IADLs

For logistic regression models, we report estimated odds ratios for each unit increase in a given covariate (Table 4). For the ordinal logistic models, the parameters are interpretable as odds ratios too, as follows. For example, for the model for general health perception in year t, the estimated odds ratio for general health perception in prior year t - 1 is 2.79 (95% confidence interval, 2.39-3.26). To interpret this quantity, suppose that general health perception at year t was dichotomized to either "good," "fair," or "poor" versus "excellent" or "very good" and that a logistic regression model was used for analysis. Then, the odds ratio for general health perception at t - 1 would express the change in odds of a "good," "fair," or "poor" response at t for each unit change in general health perception at t - 1. In the ordinal logistic regression model, this odds ratio is assumed to be constant regardless of where the response is dichotomized. That is, the odds ratio is the same as if we had dichotomized the response to be "poor" versus "fair" or better.

The previous year's health status was the most powerful predictor of general health perception, ADLs, and IADLs. Comorbidities such as the Charlson score and diabetes were also frequently important, as was increasing age. Female gender was associated with more deficiencies in ADLs and IADLs, whereas blacks were less likely to become institutionalized. In each of the five models in Table 4, we tested the interaction between age and sex and a quadratic age term, neither of which was significant. Nor did including these additional terms appreciably change the other regression co-efficients.

Discussion

Heart failure is a highly morbid syndrome among older adults. (54), (55) In our cohort, many patients had poor health status at baseline and 18% of the patients died over 1 year. Health perception and deficiencies in ADLs and IADLs were strong correlates of mortality. We found that overall health perception appeared to be more variable in health transitions than ADLs and IADLs, which were more stable. A patient's rating of his or her health perception is more subjective than ADLs or IADLs, so this mutability might not be surprising. Despite this subjectivity, health perception is clearly important as seen by the correlation with mortality.

Our study also offers important new information on several controversial or understudied areas related to quality of life in patients with heart failure. We found that women were more likely than men to have deficiencies in ADLs and IADLs but that general health perception was not associated with gender in multivariate analyses. Prior results from single-site and trial populations have conflicted regarding whether gender is associated with quality of life in patients with heart failure. (34), (56-58) Few studies have investigated the reasons for gender differences in quality of life among patients with heart failure, but it appears that suboptimal medical care in women may be one of the contributing causes. (34), (59)

In addition, no differences between blacks and whites existed for the health perception, ADLs, IADLs, and mortality outcomes. However, blacks with heart failure were less likely to be institutionalized than whites. These findings regarding the association between race and institutionalization are consistent with results from general populations of older adults. (60) The reasons for these racial disparities are complex, but probably include the structure and availability of informal caregiving, lower opportunity costs of informal caregiving time for black families, differing cultural attitudes about institutional care, provision of different styles of medical care, and fewer nursing homes in black communities. (61)

Our study has several limitations. First, we rely on Medicare claims diagnostic codes for the identification of patients with heart failure as opposed to clinical criteria. (62), (63) However, we do have a relatively stringent algorithm for a heart failure case, (41) probably biasing toward a sicker population that seeks care. Second, we lost patients to follow-up in this very sick cohort. We know little about the reasons for dropout in the MCBS, but given that the patients lost to follow-up appear to be sicker, we may be underestimating transitions to lower health and functional status in this population. Our sensitivity analyses suggested that small but not imperceptible biases could result from the dropouts.

Despite these limitations, though, our study has a number of important strengths. First, to our knowledge, we used the largest nationally representative sample of older patients with heart failure that both contains health-related quality of life information and reflects real-world practice. Most prior studies of health-related quality of life in patients with heart failure draw on highly selected trial populations or else are limited to a single or a few sites. Differences in population characteristics in these smaller studies can lead to divergent results that are difficult to interpret and apply. Second, we analyzed a variety of clinically and practically important outcomes for the older population with heart failure. These measures include mortality, institutionalization, everyday functioning assessed with ADLs and IADLs, and patients' perceptions of their general health status. Third, our statistical methods allow us to present descriptive and correlational information in several useful ways. The transition matrices present aggregate data on the probability of patients' health status improving, worsening, or remaining stable given a certain starting baseline health state. The multivariable models identify and quantify the most important factors affecting patient outcomes. Fourth, as the population ages, heart failure increasingly becomes a critical syndrome for the geriatric population. Function and overall health status are particularly important endpoints for the geriatric population, and our study provides national population-based data that could be useful for public health planning purposes of large providers, managed care organizations, and state and federal governments.

Conclusion

Overall, our study indicates that mortality, poor health-related quality of life, and declines in functioning are common among older patients with heart failure. These data are most useful for aggregate policy purposes. Application of these exact transition probabilities to individual patients would probably not be sufficiently accurate for clinical usage. However, in assessing older patients with heart failure, it is helpful to know that dramatic changes in health status over 1-year periods are relatively uncommon, and that health perception and functional status are strongly associated with mortality.

Key Points

* Among older patients with heart failure, 58% rated their general health as "fair" or "poor" and 18% died within 1 year.

* Health perception, activities of daily living, and instrumental activities of daily living were strong correlates of mortality.

* Dramatic changes in health status were relatively uncommon over 1 year among survivors.

* Decline was common in patients with "excellent" or "very good" health perception.

Technical Appendix

Statistical Analysis

Development of Transition Matrices. Our statistical analysis focuses on modeling health state transitions and presenting them in transition matrices expressing the probability of each possible state at year t for each possible state in year t - 1. The ordinal logistic regression model is particularly suited to this problem because health perception, deficiencies in ADLs, and deficiencies in IADLs are outcome variables with multiple ordered categories. Health perception ranges from poor to excellent, and deficiencies in ADLs could range from 0 to 5. Because preliminary analyses suggested that the ordinal logistic regression model assumptions were not satisfied for all dichotomizations of the response variables, some of the categories were combined for the analyses in this article. We collapsed the ADL variable into 0, 1, 2, 3, or 4+ activity deficiencies and the IADL variable into 0, 1, 2, 3, 4, or 5+ deficiencies. Health perception has five categories, which we maintained. We used the parameter estimates from the models to calculate the probability of transitioning from one given health state to another, probabilities that make up the entries in the transition matrices.

For flexible estimation of transition matrices, both linear and squared terms of each predictor at t - 1 (general health perception, ADLs, and IADLs) were included in all models, the models were adjusted for age and age-squared, and all terms were interacted with sex. Results were presented for various levels of age and sex. Because including other co-variates such as education and clinical variables in the models would also require presenting matrices for different levels of these variables, we limited our covariates to age and sex. Finally, we investigated whether the outcome probabilities varied by year (1992, 1993, and 1994), after adjusting for the predictor at t - 1, age, and sex, and found no such effects.

To appropriately model IADLs in the presence of potential institutionalization required special treatment of that variable. To handle this, we only modeled outcomes for subjects who were not institutionalized in year t - 1 and who therefore had a valid IADL measure in that year. As for the other outcomes, we modeled death in year t in a logistic model with IADLs in year t - 1 as the predictor. Then, conditional on survival, we modeled entry into an institution in year t in another logistic model. Finally, conditional on survival and no institutionalization, we modeled IADLs in year t in an ordinal logistic model. The estimated transition matrices include institutionalization as one possible outcome. For purposes of these analyses, once a subject was institutionalized, he or she was considered to remain institutionalized until the end of the study, dropout, or death.

Because we were interested in yearly transitions in health status, we used data for each 2-year period from each eligible patient. For example, a patient who answered surveys between 1991 and 1993 who then died in 1994 would contribute data to the model for the transitions in health perception and ADLs twice (1991-1992 and 1992-1993), and three times for the death analyses (1991-1992, 1992-1993, and 1993-1994). If he or she entered an institution during the 1992 to 1993 interval, he or she would also contribute 2 years of data to the model for institutionalization and 1 year of data (1991-1992) to the model for the IADL outcome. If he or she entered the institution in the 1991 to 1992 interval or was institutionalized during the baseline year, he or she would contribute no data to the model for the IADL outcome.

Independent Correlates of Health Perception and HRQOL. To account for the fact that some subjects contribute more than 1 year of data to estimation of the model, confidence intervals were constructed using robust standard errors for clustered data. (52) The models that include IADLs as a predictor only apply to those subjects not in an institution in the current year, and the IADL model is only for subjects who are not institutionalized 1 year later.

Patients Lost to Follow-up. Some subjects in our cohort are known to survive a given year but are lost to follow-up with respect to the MCBS survey. The survey includes the general health and functional status variables, and allows construction of the clinical and institutionalization variables. Vital status is known even for the subjects lost to follow-up because death information is obtained in the MCBS by means of an ongoing National Death Index search. (64) In our data set, 9% of 1-year follow-up observations where death did not occur are missing. We investigated the possible biases in the fitted transition matrices that could arise due to this "missingness" by looking at the previous year's outcomes as predictors of being lost to follow-up and by performing sensitivity analyses.

Table 1. Baseline study sample, MCBS heart failure cohort, 1991 (a), (b)

                                                         Total (n = 872)

Sociodemographics

  Age (yr) (c)

  Mean (SE)                                              79.2 (0.28)

  Minimum, maximum                                          65, 103

    Female (%)                                           561 (62)

    Race (%)

      White                                              743 (86)

      Black                                               89 (9)

      Other/unknown                                       40 (5)

  Education (yr) (d) (%)

    0-8                                                  318 (37)

    9-11                                                 130 (17)

    12                                                   208 (28)

    13+                                                  136 (18)

Medical history (%)

  Hypertension                                           370 (43)

  Coronary artery disease                                484 (57)

  Myocardial infarction                                   93 (12)

  Diabetes                                               231 (28)

  Charlson Comorbidity Index

    0                                                    316 (35)

    1                                                    243 (28)

    2-3                                                  226 (27)

    4-10                                                  87 (11)

Baseline health perceptions and functional status (%)

  Health Perception
    Excellent                                             47 (5)

    Very good                                            105 (12)

    Good                                                 226 (25)

    Fair                                                 284 (33)

    Poor                                                 210 (25)

  Deficiencies in activities of daily living (%)

    0                                                    358 (45)

    1                                                    139 (16)

    2                                                    101 (11)

    3                                                     77 (9)

    4                                                    115 (11)

    5                                                     82 (8)

  Deficiencies in instrumental activities of daily
  living (e) (%)

    0                                                    182 (27)

    1                                                    144 (20)

    2                                                    102 (14)

    3                                                     75 (10)

    4                                                     86 (12)

    5                                                     71 (9)

    6                                                     60 (7)

                                                         Men (n = 311)

Sociodemographics

  Age (yr) (c)

  Mean (SE)                                              77.2 (0.43)

  Minimum, maximum                                          66, 103

    Female (%)

    Race (%)

      White                                              269 (87)

      Black                                               26 (8)

      Other/unknown                                       16 (6)

  Education (yr) (d) (%)

    0-8                                                  110 (35)

    9-11                                                  49 (18)

    12                                                    75 (28)

    13+                                                   54 (20)

Medical history (%)

  Hypertension                                           113 (37)

  Coronary artery disease                                202 (67)

  Myocardial infarction                                   52 (17)

  Diabetes                                                82 (28)

  Charlson Comorbidity Index

    0                                                     92 (30)

    1                                                     86 (27)

    2-3                                                   99 (31)

    4-10                                                  34 (11)

Baseline health perceptions and functional status (%)

  Health Perception

    Excellent                                             20 (5)

    Very good                                             40 (13)

    Good                                                  87 (28)

    Fair                                                  83 (27)

    Poor                                                  81 (27)

  Deficiencies in activities of daily living (%)

    0                                                    165 (56)

    1                                                     45 (15)

    2                                                     25 (7)

    3                                                     20 (6)

    4                                                     31 (8)

    5                                                     25 (7)

  Deficiencies in instrumental activities of daily
  living (e) (%)

    0                                                    109 (41)

    1                                                     59 (22)

    2                                                     23 (8)

    3                                                     20 (7)

    4                                                     24 (9)

    5                                                     21 (7)

    6                                                     20 (7)

                                                         Women (n = 561)

Sociodemographics

  Age (yr) (c)

  Mean (SE)                                              80.4 (0.33)

  Minimum, maximum                                          65, 103

    Female (%)

    Race (%)

      White                                              474 (85)

      Black                                               63 (10)

      Other/unknown                                       24 (5)

  Education (yr) (d) (%)

    0-8                                                  208 (38)

    9-11                                                  81 (17)

    12                                                   133 (29)

    13+                                                   82 (16)

Medical history (%)

  Hypertension                                           257 (47)

  Coronary artery disease                                282 (51)

  Myocardial infarction                                   41 (8)

  Diabetes                                               149 (29)

  Charlson Comorbidity Index

    0                                                    224 (38)

    1                                                    157 (29)

    2-3                                                  127 (23)

    4-10                                                  53 (10)

Baseline health perceptions and functional status (%)

  Health Perception

    Excellent                                             27 (4)

    Very good                                             65 (11)

    Good                                                 139 (24)

    Fair                                                 201 (36)

    Poor                                                 129 (24)

  Deficiencies in activities of daily living (%)

    0                                                    193 (38)

    1                                                     94 (16)

    2                                                     76 (14)

    3                                                     57 (10)

    4                                                     84 (13)

    5                                                     57 (9)

  Deficiencies in instrumental activities of daily
  living (e) (%)

    0                                                     73 (18)

    1                                                     85 (19)

    2                                                     79 (18)

    3                                                     55 (12)

    4                                                     62 (15)

    5                                                     50 (10)

    6                                                     40 (7)

                                                         P value

Sociodemographics

  Age (yr) (c)                                           <0.001

  Mean (SE)

  Minimum, maximum

    Female (%)

    Race (%)                                              0.55

      White

      Black

      Other/unknown

  Education (yr) (d) (%)                                  0.24

    0-8

    9-11

    12

    13+

Medical history (%)

  Hypertension                                            0.0039

  Coronary artery disease                                <0.0001

  Myocardial infarction                                   0.0013

  Diabetes                                                0.84

  Charlson Comorbidity Index                              0.013

    0

    1

    2-3

    4-10

Baseline health perceptions and functional status (%)

  Health Perception                                       0.48

    Excellent

    Very good

    Good

    Fair

    Poor

  Deficiencies in activities of daily living (%)         <0.001

    0

    1

    2

    3

    4

    5

  Deficiencies in instrumental activities of daily
  living (e) (%)                                         <0.001

    0

    1

    2

    3

    4

    5

    6

(a) MCBS, Medicare Current Beneficiary Survey.
(b) Entries are counts. Corresponding percentages are weighted to
account for sample survey design. For categorical and binary variables,
we used [chi square] tests to test for association with gender,
accounting for the sample survey design. For the ordinal
variables such as education or general health perception,
we used a two-sample t test, again accounting for the sample design.
(c) Age at baseline (1991).
(d) 80 subjects are missing education.
(e) 152 subjects are missing IADLs due to institutionalization at
baseline.

Table 2. Cumulative natural history, MCBS heart failure cohort,
1991-1994 (a), (b)

                                1991      1992      1993      1994
                                (%)       (%)       (%)       (%)

Living in community           720 (83)  462 (53)  344 (39)  260 (30)
Institutionalized (ever) (c)  152 (17)  150 (17)  121 (14)   93 (11)
Lost to follow-up                        95 (11)  121 (14)  142 (16)
Died                                    165 (19)  286 (33)  377 (43)
Total                             872       872       872       872

(a) MCBS, Medicare Current Beneficiary Summary.
(b) Unweighted numbers.
(c) Here, institutionalized indicates subjects who are alive, not lost
to follow-up, and who have lived in an institution at any time during
follow-up through the reference year.

Table 3. Year-to-year transitions in general health perception, ADL
dependencies, and IADL dependencies, MCBS heart failure cohort,
1991-1994 (a), (b)
                                 [GHP.sub.t]
Sex     Age (yr)  [GHP.sub.t-1]  Excellent  Very good  Good   Fair

Male       70     Excellent        19 (e)     30       29      11
                  Very good        13         26 (e)   35      16
                  Good              6         16       35 (e)  29
                  Fair              2          5       20      43 (e)
                  Poor              0          1        5      27
           80     Excellent        19 (e)     30       27      10
                  Very good        13         26 (e)   33      15
                  Good              6         16       34 (e)  27
                  Fair              2          5       20      40 (e)
                  Poor              0          1        5      25
Female     70     Excellent        30 (e)     35       24       7
                  Very good        12         26 (e)   36      18
                  Good              4         13       34 (e)  35
                  Fair              1          5       20      45 (e)
                  Poor              0          2        8      37
           80     Excellent        30 (e)     33       22       6
                  Very good        13         26 (e)   34      16
                  Good              5         13       32 (e)  31
                  Fair              1          5       19      41 (e)
                  Poor              0          2        7      33

                                 [ADL.sub.t]
Sex     Age (yr)  [ADL.sub.t-1]    0       1       2       3

Male       70           0        75 (e)   6       3       1
                        1        55      15 (e)   8       5
                        2        28      18      16 (e)  12
                        3        11      11      15      18 (e)
                        4         3       4       8      14
           80           0        68 (e)   9       4       2
                        1        44      18 (e)  11       7
                        2        20      16      17 (e)  15
                        3         7       8      13      17 (e)
                        4         2       3       5      11
Female     70           0        78 (e)   9       4       2
                        1        55      17 (e)  10       6
                        2        30      20      17 (e)  13
                        3        13      13      17      19 (e)
                        4         5       6      11      17
           80           0        65 (e)  13       6       3
                        1        40      19 (e)  14       9
                        2        18      16      17 (e)  16
                        3         7       8      13      18 (e)
                        4         3       3       6      12

                                 [GHP.sub.t]
Sex     Age (yr)  [GHP.sub.t-1]  Poor   D (d)

Male      70       Excellent      2       9
                   Very good      3       8
                   Good           6       9
                   Fair          17      13
                   Poor          40 (e)  27
           80      Excellent      1      13
                   Very good      2      10
                   Good           5      12
                   Fair          15      18
                   Poor          34 (e)  35
Female     70      Excellent      1       4
                   Very good      3       5
                   Good           8       6
                   Fair          20       9
                   Poor          40 (e)  13
           80      Excellent      1       8
                   Very good      2      10
                   Good           7      13
                   Fair          17      18
                   Poor          33 (e)  25

                                 [ADL.sub.t]
Sex     Age (yr)  [ADL.sub.t-1]    4     D (d)

Male       70           0         1      13
                        1         3      13
                        2        10      16
                        3        24      21
                        4        40 (e)  30
           80           0         1      15
                        1         5      15
                        2        15      17
                        3        32      23
                        4        46 (e)  33
Female     70           0         1       6
                        1         4       7
                        2        11      10
                        3        24      14
                        4        40 (e)  21
           80           0         2      10
                        1         7      12
                        2        17      15
                        3        32      22
                        4        44 (e)  32

(a) [GHP.sub.v] general health perception in year t; [ADL.sub.v]
activities of daily living dependencies in year t; [IADL.sub.v]
instrumental activities of daily living dependencies for
noninstitutionalized population in year t; t, 1992, 1993, 1994;
MCBS, Medicare Current Beneficiary Survey.
(b) Entries are estimated probabilities (x 100) of transitioning from
t - 1 state to t state in 1 year.
(c) I, transition to living in an institution in year t. This is treated
as an absorbing state in this table.
(d) D, death in year t.
(e) Indicates probability of remaining in the same category from year t
- 1 to year t.
                                          [IADL.sub.t]
Sex     Age (yr)  [IADL.sub.t-1]    0       1       2       3

Male       70           0         63 (e)  16       3       1
                        1         35      31 (e)   9       4
                        2         15      29      15 (e)   8
                        3          6      18      15      12 (e)
                        4          3       9      10      10
                        5          1       5       6       7
           80           0         55 (e)  21       4       2
                        1         27      31 (e)  11       5
                        2         10      24      15 (e)  10
                        3          4      13      13      11 (e)
                        4          2       6       8       9
                        5          1       3       4       6
Female     70           0         56 (e)  27       6       2
                        1         33      35 (e)  11       5
                        2         17      32      17 (e)  10
                        3          7      21      17      13 (e)
                        4          3      11      12      13
                        5          1       5       6       8
           80           0         44 (e)  30       7       3
                        1         24      34 (e)  13       7
                        2         11      26      17 (e)  11
                        3          5      15      15      13 (e)
                        4          2       7       9      10
                        5          1       3       4       6

                                           [IADL.sub.t]
Sex     Age (yr)  [IADL.sub.t-1]    4       5     I (c)  D (d)

Male       70           0          1       0       4      12
                        1          3       2       5      12
                        2          8       5       7      13
                        3         15      10       8      16
                        4         18 (e)  18       9      22
                        5         17      23 (e)   9      32
           80           0          1       1       6      11
                        1          4       2       8      11
                        2         11       6      11      13
                        3         17      14      13      15
                        4         18 (e)  22      14      21
                        5         15      27 (e)  14      31
Female     70           0          2      1        2       5
                        1          4      2        2       7
                        2          9      5        2       8
                        3         17      11       2      11
                        4         22 (e)  22       2      15
                        5         21      35 (e)   2      21
           80           0          3       1       7       5
                        1          6       3       7       7
                        2         12       7       7       9
                        3         19      15       8      12
                        4         21 (e)  27       8      16
                        5         17      39(e)    9      21

Table 4. Independent associations with health perception, ADLs, IADLs,
institutionalization, and death (odds ratio, 95% confidence interval)
(a), (b)

                               Health
                             perception              ADLs

Sociodemographic

  Female                  1.07 (0.86-1.32)      1.33 (1.05-1.69) (c)
  Age (per 10 yr)         0.99 (0.98-1.00)      1.07 (1.05-1.09) (c)
  Race
    Black                 0.90 (0.70-1.16)      0.98 (0.72-1.33)
    Other                 1.18 (0.81-1.73)      0.80 (0.48-1.34)
  Education (yr)
    0-8                   1.41 (1.09-1.82) (c)  1.04 (0.76-1.43)
    9-11                  1.09 (0.81-1.47)      1.02 (0.73-1.42)
    13+                   0.92 (0.69-1.23)      0.83 (0.58-1.20)

Clinical

   Hypertension           1.06 (0.88-1.26)      0.88 (0.71-1.08)
   Myocardial infarction  0.98 (0.70-1.37)      1.21 (0.88-1.65)
   Diabetes               1.25 (0.99-1.59)      1.79 (1.40-2.28) (c)
   Charlson Comorbidity
     Index
     1                    0.99 (0.77-1.28)      1.03 (0.80-1.33)
     2-3                  1.15 (0.90-1.47)      1.31 (0.99-1.74)
     4-10                 1.68 (1.19-2.37) (c)  2.01 (1.48-2.74) (c)

Prior year health status

   Health Perception      2.79 (2.39-3.26) (c)  -
   ADLs                   -                     2.97 (2.61-3.39) (c)
   IADLs                  -                     -

                                IADLs           Institutionalization

Sociodemographic

  Female                  1.53 (1.20-1.95) (c)  0.71 (0.42-1.20)
  Age (per 10 yr)         1.06 (1.04-1.08) (c)  1.07 (1.03-1.10) (c)
  Race
    Black                 1.17 (0.82-1.68)      0.38 (0.17-0.84) (c)
    Other                 1.27 (0.93-1.75)      0.27 (0.08-0.98) (c)
  Education (yr)
    0-8                   1.54 (1.16-2.06) (c)  0.65 (0.37-1.17)
    9-11                  1.51 (1.03-2.21) (c)  0.86 (0.45-1.66)
    13+                   1.31 (0.94-1.81)      0.59 (0.31-1.12)

Clinical

   Hypertension           1.01 (0.80-1.26)      1.30 (0.86-1.97)
   Myocardial infarction  0.88 (0.59-1.30)      1.09 (0.56-2.11)
   Diabetes               1.42 (1.11-1.81) (c)  1.36 (0.84-2.22)
   Charlson Comorbidity
     Index
     1                    1.32 (1.00-1.75)      0.61 (0.33-1.11)
     2-3                  1.35 (0.98-1.86)      1.23 (0.77-1.98)
     4-10                 1.70 (1.16-2.48) (c)  0.81 (0.38-1.72)

Prior year health status

   Health Perception      -                     0.99 (0.80-1.24)
   ADLs                   -                     1.10 (0.90-1.34)
   IADLs                  2.46 (2.22-2.73) (c)  1.16 (0.97-1.39)

                                Death

Sociodemographic

  Female                  0.73 (0.49-1.10)
  Age (per 10 yr)         1.04 (1.01-1.07) (c)
  Race
    Black                 1.13 (0.71-1.82)
    Other                 0.68 (0.30-1.51)
  Education (yr)
    0-8                   0.64 (0.43-0.94) (c)
    9-11                  0.63 (0.36-1.10)
    13+                   1.20 (0.76-1.89)

Clinical

   Hypertension           0.68 (0.47-0.96) (c)
   Myocardial infarction  2.52 (1.69-3.76) (c)
   Diabetes               1.19 (0.86-1.65)
   Charlson Comorbidity
     Index
     1                    1.32 (0.83-2.12)
     2-3                  1.70 (1.12-2.59) (c)
     4-10                 3.41 (2.13-5.46) (c)

Prior year health status

   Health Perception      1.26 (1.06-1.49) (c)
   ADLs                   1.09 (0.94-1.25)
   IADLs                  1.18 (1.00-1.38)

(a) ADLs, activities of daily living; IADLs, instrumental activities of
daily living.
(b) Some subjects were missing the education variable so sample sizes
are slightly smaller than those computable from Table 2. The greatest
"missingness" percent was 6%, occurring for the models for general
health perception and ADLs. For the other models, the education was
missing for less than 2% of subjects. Reference categories: Race =
white, Education = 12 years, Charlson Comorbidity Index = 0.
Categories: Health Perception (excellent, very good, good, fair,
poor), ADL deficiencies (0, 1, 2, 3, 4+) and IADL deficiencies (0, 1,
 2, 3, 4, 5+).
(c) P [less than or equal to] 0.05.

Acknowledgment 

We thank Naoko Muramatsu, PhD, for her helpful comments on an earlier draft of the article.

From the Section of General Internal Medicine, Department of Medicine, and Department of Health Studies, University of Chicago, Chicago, IL; and Department of Community and Family Medicine, the Chinese University of Hong Kong, Hong Kong.

Supported by National Institute on Aging Grant R03 AG17200-01 and American Heart Association, Midwest Affiliate, Grant-in-Aid 9951427Z. Dr. Chin is a Robert Wood Johnson Foundation Generalist Physician Faculty Scholar.

Reprint requests to Marshall H. Chin, MD, MPH, University of Chicago, 5841 S. Maryland Avenue, MC 2007, Chicago, IL 60637. Email: [e-mail]

Accepted May 1, 2003.

Copyright [c] 2003 by The Southern Medical Association

0038-4348/03/9611-1096

References

1. Hunt SA, Baker DW, Chin MH, et al. ACC/AHA guidelines for the evaluation and management of chronic heart failure in the adult: A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Revise the 1995 Guidelines for the Evaluation and Management of Heart Failure). J Am Coll Cardiol 2001;38:2101-2113.

2. U.S. Department of Health and Human Services. Health Care Financing Review: Statistical Supplement. Bethesda, MD, U.S. Department of Health and Human Services, 1995, p 56.

3. Ho KK, Anderson KM, Kannel WB, et al. Survival after the onset of congestive heart failure in Framingham Heart Study subjects. Circulation 1993;88:107-115.

4. Krumholz HM, Parent EM, Tu N, et al. Readmission after hospitalization for congestive heart failure among Medicare beneficiaries. Arch Intern Med 1997;157:99-104.

5. Vinson JM, Rich MW, Sperry JC, et al. Early readmission of elderly patients with congestive heart failure. J Am Geriatr Soc 1990;38:1290-1295.

6. Berry C, McMurray J. A review of quality-of-life evaluations in patients with congestive heart failure. Pharmacoeconomics 1999;16:247-271.

7. Jenkinson C, Jenkinson D, Shepperd S, et al. Evaluation of treatment for congestive heart failure in patients aged 60 years and older using generic measures of healthy status (SF-36 and COOP charts). Age Ageing 1997;26:7-13.

8. Rector TS, Kubo SH, Cohn JN. Patients' self-assessment of their congestive heart failure: Part 2--Content, reliability and validity of a new measure: The Minnesota Living with Heart Failure questionnaire. Heart Failure 1987;3:198-209.

9. Feinstein AR, Fisher MB, Pigeon JG. Changes in dyspnea-fatigue ratings as indicators of quality of life in the treatment of congestive heart failure. Am J Cardiol 1989;64:50-55.

10. Guyatt GH, Nogradi S, Halcrow S, et al. Development and testing of a new measure of health status for clinical trials in heart failure. J Gen Intern Med 1998;4:101-107.

11. Jaeschke R, Singer J, Guyatt GH. Measurement of health status: Ascertaining the minimal clinically important difference. Control Clin Trials 1989;10:407-415.

12. Wiklund I, Lindvall K, Swedberg K, et al. Self-assessment of quality of life in severe heart failure: An instrument for clinical use. Scand J Psychol 1987;28:220-225.

13. Havranek EP, McGovern KM, Weinberger J, et al. Patient preferences for heart failure treatment: Utilities are valid measures of health-related quality of life in heart failure. J Card Fail 1999;5:85-91.

14. Wyrwich KW, Nienaber NA, Tierney WM, et al. Linking clinical relevance and statistical significance in evaluating intra-individual changes in health-related quality of life. Med Care 1999;37:469-478.

15. O'Keeffe ST, Lye M, Donnellan C, et al. Reproducibility and responsiveness of quality of life assessment and six minute walk test in elderly heart failure patients. Heart 1998;80:377-382.

16. Green CP, Porter CB, Bresnahan DR, et al. Development and evaluation of the Kansas City Cardiomyopathy Questionnaire: A new health status measure of heart failure. J Am Coll Cardiol 2000;35:1245-1255.

17. Leidy NK, Rentz AM, Zycynski TM. Evaluating health-related quality-of-life outcomes in patients with congestive heart failure: A review of recent randomized controlled trials. Pharmacoeconomics 1999;15:19-46.

18. Rogers WJ, Johnstone DE, Yusuf S, et al. Quality of life among 5,025 patients with left ventricular dysfunction randomized between placebo and enalapril: The Studies of Left Ventricular Dysfunction. J Am Coll Cardiol 1994;23:393-400.

19. Kubo SH, Gollub S, Bourge R, et al. Beneficial effects of pimobendan on exercise tolerance and quality of life in patients with heart failure: Results of a multicenter trial. Circulation 1992;85:942-949.

20. Blackwood R, Mayou RA, Garnham JC, et al. Exercise capacity and quality of life in the treatment of heart failure. Clin Pharmacol Ther 1990;48:325-332.

21. Tandon PK, Stander H, Scharz RP Jr. Analysis of quality of life data from a randomized, placebo-controlled heart-failure trial. J Clin Epidemiol 1989;42:955-962.

22. Kostis JB, Rosen RC, Cosgrove NM, et al. Nonpharmacologic therapy improves functional and emotional status in congestive heart failure. Chest 1994;106:996-1001.

23. Cowley AJ, Skene AM. Treatment of severe heart failure: Quantity or quality of life? A trial of enoximone. Br Heart J 1994;72:226-230.

24. Rich MW, Beckham V, Wittenberg C, et al. A multidisciplinary intervention to prevent the readmission of elderly patients with congestive heart failure. N Engl J Med 1995;333:1190-1195.

25. Hawthorne MH, Hixon ME. Functional status, mood disturbance and quality of life in patients with heart failure. Prog Cardiovasc Nurs 1994;9:22-32.

26. Quittan M, Sturm B, Wiesinger GF, et al. Quality of life in patients with chronic heart failure: A randomized controlled trial of changes induced by a regular exercise program. Scand J Rehabil Med 1999;31:223-228.

27. Kavanagh T, Myers MG, Baigrie RS, et al. Quality of life and cardiorespiratory function in chronic heart failure: Effects of 12 months' aerobic training. Heart 1996;76:42-49.

28. Gorkin L, Norvell NK, Rosen RC, et al. Assessment of quality of life as observed from the baseline data of the Studies of Left Ventricular Dysfunction (SOLVD) trial quality-of-life substudy. Am J Cardiol 1993;71:1069-1073.

29. Stewart AL, Greenfield S, Hays RD, et al. Functional status and well-being of patients with chronic conditions: Results from the medical outcomes study. JAMA 1989;262:907-913.

30. Albanese MC, Plewka M, Gregori D, et al. Use of medical resources and quality of life of patients with chronic heart failure: A prospective survey in a large Italian community hospital. Eur J Heart Fail 1999;1:411-417.

31. Wolinsky FD, Wyrwich KW, Nienaber NA, et al. Generic versus disease-specific health status measures: An example using coronary artery disease and congestive heart failure patients. Eval Health Prof 1998;21:216-243.

32. Ware JE Jr, Snow KK, Kosinski J, et al. SF-36 Health Survey Manual and Interpretation Guide. Boston, The Health Institute, New England Medical Center, 1993.

33. Romm FJ, Hulka BS, Mayo F. Correlates of outcomes in patients with congestive heart failure. Med Care 1976;14:765-776.

34. Chin MH, Goldman L. Gender differences in 1-year survival and quality of life among patients admitted with congestive heart failure. Med Care 1998;36:1033-1046.

35. Jaagosild P, Dawson NV, Thomas C, et al. Outcomes of acute exacerbation of severe congestive heart failure: Quality of life, resource use, and survival--SUPPORT Investigators: The Study to Understand Prognosis and Preferences for Outcomes and Risks of Treatments. Arch Intern Med 1998;158:1081-1089.

36. Konstam V, Salem D, Pouleur H, et al. Baseline quality of life as a predictor of mortality and hospitalization in 5025 patients with congestive heart failure. Am J Cardiol 1996;78:890-895.

37. Diehr P, Williamson J, Patrick DL, et al. Patterns of self-rated health in older adults before and after sentinel events. J Am Geriatr Soc 2001;49:36-44.

38. Fox E, Landrum-McNiff K, Zhong Z, et al. Evaluation of prognostic criteria for determining hospice eligibility in patients with advanced lung, heart, or liver disease. JAMA 1999;282:1638-1645.

39. Adler GS. A profile of the Medicare Current Beneficiary Survey. Health Care Financ Rev 1994;15:153-163.

40. Medicode. The International Classification of Diseases, Ninth Revision, Clinical Modification. Salt Lake City, UT, Medicode Publications, ed 5, 1995.

41. Hebert PL, Geiss LS, Tierney EF, et al. Identifying persons with diabetes using Medicare claims data. J Med Qual 1999;14:270-277.

42. Patrick DL, Erickson P. Health Status and Health Policy: Allocating Resources to Health Care. New York, Oxford University Press, 1993.

43. Mossey JM, Shapiro E. Self-rated health: A predictor of mortality among the elderly. Am J Public Health 1984;72:800-808.

44. Katz S, Ford AB, Moskowitz RW, et al. Studies of illness in the aged: The Index of ADL--A standardized measure of biological and psychosocial function. JAMA 1963;185:914-919.

45. Lawton MP, Brody EM. Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist 1969;9:179-186.

46. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613-619.

47. Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: Differing perspectives. J Clin Epidemiol 1993;46:1075-1079.

48. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 1987;40:373-383.

49. Chin MH, Goldman L. Correlates of early hospital readmission or death in patients with congestive heart failure. Am J Cardiol 1997;79:1640-1644.

50. Diehr P, Patrick DL. Probabilities of transition among health states for older adults. Qual Life Res 2001;10:431-442.

51. McCullagh P. Regression models for ordinal data. J R Stat Soc [Ser B] 1980;42:109-142.

52. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986;42:121-130.

53. Stata Corp. Stata Statistical Software: Release 7.0. College Station, TX, Stata Corp., 2001.

54. Havranek EP, Masoudi FA, Westfall KA, et al. Spectrum of heart failure in older patients: Results from the national heart failure project. Am Heart J 2002;143:412-417.

55. Pulignano G, Del Sindaco D, Tavazzi L, et al. Clinical features and outcomes of elderly outpatients with heart failure followed up in hospital cardiology units: Data from a large nationwide cardiology database (IN-CHF Registry). Am Heart J 2002;143:45-55.

56. Riedinger MS, Dracup KA, Brecht ML, et al. Quality of life in patients with heart failure: Do gender differences exist? Heart Lung 2001;30:105-116.

57. Cline CM, Willenheimer RB, Erhardt LR, et al. Health-related quality of life in elderly patients with heart failure. Scand Cardiovasc J 1999;33:278-285.

58. Riegel B, Moser DK, Carlson B, et al. Gender differences in quality of life are minimal in patients with heart failure. J Card Fail 2003;9:42-48.

59. Riedinger MS, Dracup KA, Brecht ML. Predictors of quality of life in women with heart failure. J Heart Lung Transplant 2000;19:598-608.

60. Miller B, McFall S, Campbell RT. Changes in sources of community long-term care among African American and white frail older persons. J Gerontol 1994;49:S14-S24.

61. Cagney KA, Agree EM. Racial differences in skilled nursing care and home health use: The mediating effects of family structure and social class. J Gerontol B Psychol Sci Soc Sci 1999;54B:S223-S236.

62. Jollis JG, Ancukiewicz M, DeLong ER, et al. Discordance of databases designed for claims payment versus clinical information systems: Implications for outcomes research. Ann Intern Med 1993;119:844-850.

63. Psaty BM, Boineau R, Kuller LH, et al. The potential costs of upcoding for heart failure in the United States. Am J Cardiol 1999;84:108-109.

64. U.S. Department of Health and Human Services. National Death Index User's Manual. Hyattsville, MD, U.S. Department of Health and Human Services, 2000.

[ILLUSTRATION OMITTED]

Marshall H. Chin, MD, MPH, James X. Zhang, PHD, MS, and Paul J. Rathouz, PHD