Mary Beth Hamel

The Clinical Course of Advanced Dementia

BACKGROUNDnDementia is a leading cause of death in the United States but is underrecognized as a terminal illness. The clinical course of nursing home residents with advanced dementia has not been well described.nnnMETHODSnWe followed 323 nursing home residents with advanced dementia and their health care proxies for 18 months in 22 nursing homes. Data were collected to characterize the residents survival, clinical complications, symptoms, and treatments and to determine the proxies understanding of the residents prognosis and the clinical complications expected in patients with advanced dementia.nnnRESULTSnOver a period of 18 months, 54.8% of the residents died. The probability of pneumonia was 41.1%; a febrile episode, 52.6%; and an eating problem, 85.8%. After adjustment for age, sex, and disease duration, the 6-month mortality rate for residents who had pneumonia was 46.7%; a febrile episode, 44.5%; and an eating problem, 38.6%. Distressing symptoms, including dyspnea (46.0%) and pain (39.1%), were common. In the last 3 months of life, 40.7% of residents underwent at least one burdensome intervention (hospitalization, emergency room visit, parenteral therapy, or tube feeding). Residents whose proxies had an understanding of the poor prognosis and clinical complications expected in advanced dementia were much less likely to have burdensome interventions in the last 3 months of life than were residents whose proxies did not have this understanding (adjusted odds ratio, 0.12; 95% confidence interval, 0.04 to 0.37).nnnCONCLUSIONSnPneumonia, febrile episodes, and eating problems are frequent complications in patients with advanced dementia, and these complications are associated with high 6-month mortality rates. Distressing symptoms and burdensome interventions are also common among such patients. Patients with health care proxies who have an understanding of the prognosis and clinical course are likely to receive less aggressive care near the end of life.

Resuscitation Preferences Among Patients With Severe Congestive Heart Failure Results From the SUPPORT Project

BACKGROUNDnWe sought to describe the resuscitation preferences of patients hospitalized with an exacerbation of severe congestive heart failure, perceptions of those preferences by their physicians, and the stability of the preferences.nnnMETHODS AND RESULTSnOf 936 patients in this study, 215 (23%) explicitly stated that they did not want to be resuscitated. Significant correlates of not wanting to be resuscitated included older age, perception of a worse prognosis, poorer functional status, and higher income. The physicians perception of the patients preference disagreed with the patients actual preference in 24% of the cases overall. Only 25% of the patients reported discussing resuscitation preferences with their physician, but discussion of preferences was not significantly associated with higher agreement between the patient and physician. Of the 600 patients who responded to the resuscitation question again 2 months later, 19% had changed their preferences, including 14% of those who initially wanted resuscitation (69 of 480) and 40% of those who initially did not (48 of 120). The physicians perception of the patients hospital resuscitation preference was correct for 84% of patients who had a stable preference and 68% of those who did not.nnnCONCLUSIONSnAlmost one quarter of patients hospitalized with severe heart failure expressed a preference not to be resuscitated. The physicians perception of the patients preference was not accurate in about one quarter of the cases. but communication was not associated with greater agreement between the patient and the physician. A substantial proportion of patients who did not want to be resuscitated changed their minds within 2 months of discharge.

Patient age and decisions to withhold life-sustaining treatments from seriously ill, hospitalized adults

For patients hospitalized with serious illness, decisions about the use of invasive, life-sustaining treatments are medically and ethically complex (1-3). Previous research suggests that treatment decisions may be based on patient age, independent of medical appropriateness or patients preferences (4, 5). We previously demonstrated that seriously ill elderly patients receive fewer procedures and less expensive hospital care than younger patients with similar illnesses. This preferential allocation of hospital services to younger patients does not seem to be driven by differences in patients preferences for life-extending care or illness characteristics (4). Higher rates of withholding life-sustaining treatments from elderly persons could contribute to these observed differences in resource use. Controversy exists about whether patient age is an acceptable criterion on which to base decisions about use of health care resources. Some researchers have argued that withholding beneficial treatments from elderly patients to make more health care resources available to younger patients is rational and inherently fair (6-10). Others have argued that age is an inappropriate and arbitrary criterion by which to allocate health care resources and that treatment decisions should be based on a patients ability to benefit (11-14). We studied 9105 seriously ill patients to evaluate how patient age, independent of patients objective prognoses and preferences for life-extending care, influences physicians decisions to withhold life-sustaining therapies. Methods Study Design The study group consisted of patients enrolled in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT), a study of preferences, decision making, and outcomes for patients hospitalized at one of five geographically diverse academic medical centers. A full description of the methods and objectives of this study has been published elsewhere (15, 16). Patients were screened for eligibility at hospital admission; patients in intensive care units were also screened daily. Patients were enrolled if they were 18 years of age or older and met illness severity criteria for at least one of nine diagnostic categories: acute respiratory failure, chronic obstructive lung disease, congestive heart failure, cirrhosis, nontraumatic coma, metastatic colon cancer, advanced lung cancer, multiple organ system failure with sepsis, or multiple organ system failure with malignancy. Diagnostic criteria were designed so that patients would have, on average, a 50% probability of surviving for 6 months. Eligible patients who died or were discharged within 48 hours were excluded. On the basis of problems in care identified during the observational portion of the study, phase I (enrollment from June 1989 through June 1991), an intervention was developed and implemented during phase II (enrollment from January 1992 through January 1994). Clinicians of patients who were randomly assigned to the intervention group received information about their patients preferences and prognoses and were assigned clinical nurse-specialists to facilitate symptom control and communication with their patients. The study design was approved by the institutional review board at each medical center, and informed consent was obtained before patients were interviewed. Because no differences in targeted outcomes were observed between the intervention group and the control group in phase II or between phase I and phase II patients (16), we combined phase I and phase II patients for these analyses. Data Collection Data were collected daily from paper and electronic medical records and from interviews with patients and their surrogates (defined as persons who would make care decisions if patients were unable to do so). Patients were excluded from interviews if they could not communicate because of such reasons as intubation, coma, or cognitive impairment. Abstractors of medical records gathered information, including diagnoses, comorbid conditions, and acute physiology data that were included in the Acute Physiology and Chronic Health Evaluation (APACHE) III prognostic system (17). In addition, abstractors collected information on the presence and timing of discussions and decisions about use of ventilators, surgery, dialysis, blood transfusions, vasopressors, organ transplantation, tube feeding, cardiopulmonary resuscitation, and treatment in intensive care units. Surgery was defined as a procedure that took place in the operating room. Between study days 2 and 6, interviewers questioned patients and their surrogates about patients sociodemographic characteristics, functional and activity status before hospital admission (using a modified version of the Katz activities of daily living scale [18, 19] and the Duke Activity Status Index [20, 21]), preferences for cardiopulmonary resuscitation in the event of cardiac arrest, and preferences for care aimed at extending life. The exact wording of questions about patients care preferences is shown in the Appendix. Between study days 2 and 6, interviewers asked physicians about their patients preferences for cardiopulmonary resuscitation and life-extending care and about the physicians own preferences if they were in their patients situations. The latter question was asked only during phase I of the study. Statistical Analysis We focused our analyses on decisions to withhold three life-sustaining treatments (dependent variables): ventilator support, surgery, and dialysis. The presence of a decision to withhold a life-sustaining treatment was defined as chart documentation of the decision to withhold the treatment if the patients condition required such a treatment to sustain life. We analyzed these three treatments because they are invasive, expensive, and commonly used in clinical practice and were discussed often enough to make analyses feasible. We used descriptive statistics to characterize the entire study sample and the subgroup of patients for whom one or more decisions were made to withhold one of the life-sustaining treatments studied. We analyzed age as a continuous variable and as a categorical variable. In analyses of age as a continuous variable, we examined age as a simple linear predictor. In addition, we used cubic spline functions (22) to avoid the assumption of linearity between age and the timing of decisions to withhold treatments. For analyses that examined age as a categorical variable, we divided patients into five age groups on the basis of decades of life and the age distribution of the sample (<50 years, 50 to 59 years, 60 to 69 years, 70 to 79 years, and 80 years). We restricted all bivariable and multivariable analyses of each treatment to the subgroup of patients for whom the treatment issue arose at some point during the index hospitalization (defined as medical record documentation of discussion with the patient and family, discussion among hospital staff, or documentation of a note or order regarding the treatment). We used chi-square tests for trend in our bivariable comparisons between patient age group and decisions to withhold a life-sustaining treatment at some point during the index hospitalization. To adjust for potential confounding factors, we performed multivariable analyses using Cox proportional-hazards models; we analyzed time from study admission to the day of first chart documentation of a decision to withhold a particular life-sustaining treatment if the patients medical condition required such a treatment to sustain life. Patients were censored when they died or were discharged from the hospital. We adjusted for sex, ethnicity, income, insurance, education, study site, number of days hospitalized before study admission, baseline functional status, number of comorbid conditions, the presence of cancer or dementia as a baseline comorbid condition, an objective estimate of 2-month survival made on study day 3 by using the SUPPORT prognostic model (23), and the patients preferences for cardiopulmonary resuscitation and life-extending care between study days 3 and 6. For these multivariable analyses, do not know responses were classified with yes responses because cardiopulmonary resuscitation is typically provided unless patients state a clear preference to forgo this treatment. When patient interview data were not available, surrogates perceptions of patients preferences were substituted, as is done in clinical practice. Patients for whom preference data were not available from patients or surrogates could not be included in these primary analyses. The SUPPORT prognostic model based estimates of survival on 11 physiologic measures recorded on study day 3, diagnosis, age, number of days in the hospital before study entry, presence of cancer, and neurologic function. Because the SUPPORT model includes age in its survival estimates, we could adjust for the independent effect of age on survival by including SUPPORT prognostic estimates in our multivariable models. In our multivariable analyses of age as a simple continuous variable, we calculated adjusted hazard ratios (and corresponding 95% CIs) associated with each decade increase in age for decisions to withhold ventilator support, dialysis, and surgery. From the Cox proportional-hazard models containing age as a continuous variable represented by using cubic spline functions, we plotted the adjusted probability of a decision to withhold each life-sustaining therapy by study day 30 against patient age. We obtained model-based estimates of time-to-event curves (24, 25) for a typical SUPPORT patient (that is, using modal or median values for covariates other than age) at ages ranging from 18 to 100 years. From each of these curves, we selected the estimated probability at day 30. From the models containing age groups, we computed adjusted relative risk (hazard ratios) for decisions to withhold treatments and corresponding

Pragmatic Trials — Guides to Better Patient Care?

Because of concerns about the real-world applicability of clinical trials and about improving the quality and value of health care, “pragmatic” trials are attracting increasing attention. These trials have both important strengths and inherent limitations.

Resource Use and Survival of Patients Hospitalized with Congestive Heart Failure: Differences in Care by Specialty of the Attending Physician

In an effort to contain costs, many health networks have adopted strategies that increase utilization of primary care physicians and limit referrals to specialty care (1). This tactic is intended to encourage less resource-intensive patient care provided by generalists (2). Previous investigations suggest that substituting generalist care for specialist care results in suboptimal patient outcomes, with consistent findings across illnesses and specialties, including rheumatoid arthritis (3) and critical care medicine (4). This relation has been of particular interest in cardiovascular disease (5). Previous investigations of the effects of specialty care among patients with acute myocardial infarction suggest that cardiologist care is more resource intensive but may result in a survival benefit (6-8). These findings were in agreement with other studies of acute cardiac illnesses, each suggesting better patient outcomes with subspecialty care (9, 10). Specialty care may also improve clinical outcomes among patients hospitalized with congestive heart failure (11, 12). To investigate the relation between physician specialty and resource utilization and outcomes among patients with an acute exacerbation of congestive heart failure in a larger, multisite study, we studied 1298 patients in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT). Methods A prospective, multicenter study, SUPPORT was intended to examine decision making and outcomes for seriously ill, hospitalized adult patients. Detailed descriptions of the studys design, sites, patient population, and data collection strategy have been published elsewhere (13-16). Patients We studied patients enrolled in SUPPORT who had a primary diagnosis of acute exacerbation of congestive heart failure and whose attending physicians were cardiologists or general internists. The study had two phases: an observation phase and an intervention phase. Because the intervention did not affect the targeted outcomes for patients with congestive heart failure (13), patients enrolled during both phases were included in our analyses. Five geographically diverse institutions participated in SUPPORT: Beth Israel Hospital, Boston, Massachusetts; University of California Los Angeles Medical Center, Los Angeles, California; Marshfield Clinic, Marshfield, Wisconsin; Duke Medical Center, Durham, North Carolina; and MetroHealth Medical Center, Cleveland, Ohio. The institutional review board at each site approved the study design, and informed consent was obtained from patients or their surrogate decision makers at study entry. Patients with congestive heart failure were included in SUPPORT if they were admitted to the hospital or were transferred to the intensive care unit [ICU] with a primary diagnosis of an acute exacerbation of congestive heart failure. One of the following specific criteria was also required for inclusion: 1) history of severe congestive heart failure at baseline (New York Heart Association class III or IV disease) and medications before admission that included two or more representatives from the diuretic, vasodilator, or angiotensin-converting enzyme [ACE] inhibitor drug classes; 2) history of New York Heart Association class IV congestive heart failure, manifested by baseline dyspnea at rest; systolic blood pressure of 100 mm Hg or less; or a history of hypotension precluding use of the medications listed above; or 3) chart documentation of congestive heart failure and left ventricular ejection fraction of 20% or less. Patients were excluded from SUPPORT if their congestive heart failure was due to valvular disease, restrictive cardiac disease, pericardial disease, iatrogenic fluid overload, or renal failure. Patients were also excluded if they were pregnant, did not speak English, were nonresident foreign nationals, were younger than 18 years of age, had been transferred from another hospital and not to an ICU, had received a diagnosis of AIDS, were hospitalized with an estimated length of stay of less than 72 hours, or died or were discharged within 48 hours of study entry. Eligible patients cared for by physicians who were not cardiologists or general internists were excluded from this analysis. Data Collection The specialty of the primary attending physician was recorded at the time of study entry. A self-administered questionnaire obtained information about physician age, ethnicity, and sex. The survey asked physicians if they planned to care for the patient after hospitalization in the observation phase only. Data on patient sociodemographic characteristics, functional status, cardiovascular history, preferences for care, and components of the Acute Physiology Score were collected by chart abstraction and interviews with patients or their surrogate decision makers. The Acute Physiology Score has been shown to predict in-hospital death; higher scores indicate increased risk (17). Chart abstracters also gathered information on discharge medications, events (death or transfer to an ICU), and therapies such as right-heart catheterization. Patient survival was estimated by using the SUPPORT prognostic model, described elsewhere (18). The SUPPORT prognostic model had good discrimination (C statistic, 0.78) for predicting 180-day survival among SUPPORT patients. Total hospital charges were obtained from hospital billing systems. Costs were estimated by using the Medicare cost-to-charge ratio for each cost center (Medicare Uniform bill, 1982 version) and were converted to 1994 U.S. dollars by using the Medical Consumer Price Index. The intensity of patient care was measured by using the average Therapeutic Intervention Scoring System (TISS) score for days 1 through 25 of hospitalization. The TISS is a chart-derived additive measure of resource intensity that assigns one point for minor interventions (such as chest physical therapy or peripheral intravenous therapy) and two to four points for more substantial interventions (such as endotracheal intubation or cardiac catheterization). Previous studies have shown this index to be a valid and reliable measure of therapeutic intensity (19). Patients were enrolled in SUPPORT from June 1989 to June 1991 and January 1992 to January 1994. Survival to 6 months after enrollment was determined from medical records and telephone follow-up. The National Death Index was used to determine survival beyond 6 months and included follow-up to 31 December 1994. The maximum follow-up time for patients in this cohort was 4.6 years. We also abstracted randomly selected charts of patients cared for by generalists for the presence of specialty consultation and whether patients were transferred to the care of a cardiologist during hospitalization. Charts were available at four of five SUPPORT sites. Trained physician or nurse chart abstracters reviewed a total of 98 charts (17.7% of the generalist cohort). Statistical Analysis Outcomes of interest were 1) hospital costs; 2) average daily TISS score; 3) right-heart catheterization during the first 7 study days; 4) coronary angiography during hospitalization; 5) whether an echocardiogram or radionuclide [multiple-gated image acquisition] scan was obtained during hospitalization, 6) continuous electrocardiographic monitoring during the first 7 study days; 7) transfer to an ICU from a floor setting; 8) use of specific medications on discharge; 9) survival censored at 30, 180, and 365 days; and 10) survival to 31 December 1994. For bivariable comparisons, we used the Fisher exact test or Wilcoxon rank-sum test. Unadjusted survival rates were estimated by using the Kaplan-Meier product-limit method. Linear regression modeling was used in analyses of cost and TISS. Logistic regression was used to model procedures, tests, events, and use of discharge medications. Cox proportional-hazards modeling was used to estimate relative hazards for death at all censoring times noted and to estimate adjusted survival. All analyses were performed by using SAS software, version 6.10 for Macintosh or version 6.12 for Windows (SAS Institute, Inc., Cary, North Carolina). For all analyses, the independent variable of interest was the specialty of the patients attending physician, dichotomized into cardiologists and general internists. All multivariable models included adjustment for patient age; sex; ethnicity; insurance; site of patient enrollment; number of dependencies in activities of daily living; history of myocardial infarction, ventricular tachycardia, or ventricular fibrillation; whether the patient had a myocardial infarction during the study hospitalization; number of comorbid conditions; SUPPORT prognostic model 2-month survival estimate on day 1; and Acute Physiology Score on day 1. Because the Acute Physiology Score may not adequately match severity of illness in patients with congestive heart failure, we also included the most abnormal values for serum sodium, serum albumin, systolic blood pressure, heart rate, and respiratory rate measured in the first 3 study days. Logistic models of medication use also included maximum recorded creatinine concentration, maximum recorded serum potassium concentration, use of each medication at admission, and heart rhythm at admission. Cox proportional-hazards models adjusted for the number of patient comorbid conditions and a linear spline function for patient age with a knot at 65 years (Appendix Table 1). Appendix Table 1. Hazard Ratios Associated with Each Variable in Models of Survival to 1 Year All models included a propensity score as an additional adjustment for factors that might modify patients access to specialty care (20, 21). The propensity score was generated by using a logistic regression model with attending specialty as the dependent variable; it represents the likelihood of having a cardiologist as the attending physician. The propensity score model contained the patient sociodemographic and disease sever

Perspectives, Preferences, Care Practices, and Outcomes Among Older and Middle-Aged Patients With Late-Stage Cancer

PURPOSEnTo evaluate relationships among physician and cancer patient survival estimates, patients perceived quality of life, care preferences, and outcomes, and how they vary across middle-aged and older patient groups.nnnPATIENTS AND METHODSnSubjects were from the Study to Understand Prognoses and Preferences for Risks of Treatments (SUPPORT) prospective cohort studied in five US teaching hospitals (from 1989 to 1994), and included 720 middle-aged (45 to 64 years) and 696 older (> or = 65 years) patients receiving care for advanced cancer. Perspectives were assessed in physician and patient/surrogate interviews; care practices and outcomes were determined from hospital records and the National Death Index. General linear models were used within age groups to obtain adjusted estimates.nnnRESULTSnAlthough most patients had treatment goals to relieve pain, treatment preferences and care practices were linked only in the older group. For older patients, preference for life-extending treatment was associated with more therapeutic interventions and more documented discussions; cardiopulmonary resuscitation (CPR) preference was linked to more therapeutic interventions and longer survival. For middle-aged patients, better perceived quality of life was associated with preferring CPR. In both groups, patients higher survival estimates were associated with preferences for life-prolonging treatment and CPR; physicians higher survival estimates were associated with patients preferences for CPR, fewer documented treatment limitation discussions about care, and actual 6-month survival. More discussions were associated with readmissions and earlier death. More aggressive care was not related to outcomes.nnnCONCLUSIONnFewer older patients preferred CPR or life-prolonging treatments. Although older patients goals for aggressive treatment were related to care, this was not so for middle-aged patients. Aggressive care was not related to prolonged life in either group.

Race, resource use, and survival in seriously III hospitalized adults

OBJECTIVE: To examine the association between patient race and hospital resource use.DESIGN: Prospective cohort study.SETTING: Five geographically diverse teaching hospitals.PATIENTS: Patients were 9,105 hospitalized adults with one of nine illnesses associated with an average 6-month mortality of 50%.MEASUREMENTS AND MAIN RESULTS: Measures of resource use included: a modified version of the Therapeutic Intervention Scoring System (TISS); performance of any of any of five procedures (operation, dialysis, pulmonary artery catheterization, endoscopy, and bronchoscopy); and hospital charges, adjusted by the Medicare cost-to-charge ratio per cost center at each participating hospital. The median patient age was 65; 79% were white, 16% African-American, 3% Hispanic, and 2% other races; 47% died within 6 months. After adjusting for other sociodemographic factors, severity of illness, functional status, and study site, African-Americans were less likely to receive any of five procedures on study day 1 and 3 (adjusted odds ratio [OR] 0.70; 95% confidence interval [CI] 0.60, 0.81). In addition, African-Americans had lower TISS scores on study day 1 and 3 (OR −1.8; 95% CI −1.3, −2.4) and lower estimated costs of hospitalization (OR −

Generalists and Oncologists Show Similar Care Practices and Outcomes for Hospitalized Late-Stage Cancer Patients

2,805; 95% CI −

Patient characteristics associated with care by a cardiologist among adults hospitalized with severe congestive heart failure

1,672, −

Identification of comatose patients at high risk for death or severe disability. SUPPORT Investigators. Understand Prognoses and Preferences for Outcomes and Risks of Treatments.

3,883). Results were similar after adjustment for patients’ preferences and physicians’ prognostic estimates. Differences in resource use were less marked after adjusting for the specialty of the attending physician but remained significant. In a subset analysis, cardiologists were less likely to care for African-Americans with congestive heart failure (p<.001), and cardiologists used more resources (p<.001). After adjustment for other sociodemographic factors, severity of illness, functional status, and study site, survival was slightly better for African-American patients (hazard ratio 0.91; 95% CI 0.84, 0.98) than for white or other race patients.CONCLUSIONS: Seriously ill African-Americans received less resource-intensive care than other patients after adjustment for other sociodemographic factors and for severity of illness. Some of these differences may be due to differential use of subspecialists. The observed differences in resource use were not associated with a survival advantage for white or other race patients.