Trends in the use of do-not-resuscitate orders in patients hospitalized with acute myocardial infarction (Worcester Heart Attack Study). Data are means; error bars indicate 95% confidence intervals.
Timing of do-not-resuscitate (DNR) orders in relation to major complications after myocardial infarction and receipt of invasive procedures (Worcester Heart Attack Study). CABG indicates coronary artery bypass grafting; PCI, percutaneous coronary intervention; asterisk, P≤.05.
Jackson EA, Yarzebski JL, Goldberg RJ, Wheeler B, Gurwitz JH, Lessard DM, Bedell SE, Gore JM. Do-Not-Resuscitate Orders in Patients Hospitalized With Acute Myocardial InfarctionThe Worcester Heart Attack Study. Arch Intern Med. 2004;164(7):776-783. doi:10.1001/archinte.164.7.776
Coronary heart disease is the leading cause of death in Americans. Despite increased interest in end-of-life care, data regarding the use of do-not-resuscitate (DNR) orders in acutely ill cardiac patients remain extremely limited. The objectives of this study were to describe use of DNR orders, treatment approaches, and hospital outcomes in patients with acute myocardial infarction.
The study sample consisted of 4621 residents hospitalized with acute myocardial infarction at all metropolitan Worcester, Mass, area hospitals in five 1-year periods from 1991 to 1999.
Significant increases in the use of DNR orders were observed during the study decade (from 16% in 1991 to 25% in 1999). The elderly, women, and patients with previous diabetes mellitus or stroke were more likely to have DNR orders. Patients with DNR orders were significantly less likely to be treated with effective cardiac medications, even if the DNR order occurred late in the hospital stay. Less than 1% of patients were noted to have DNR orders before hospital admission. Patients with DNR orders were significantly more likely to die during hospitalization than patients without DNR orders (44% vs 5%).
The results of this community-wide study suggest increased use of DNR orders in patients hospitalized with acute myocardial infarction during the past decade. Use of certain cardiac therapies and hospital outcomes are different between patients with and without DNR orders. Further efforts are needed to characterize the use of DNR orders in patients with acute coronary disease.
During the past 2 decades, emphasis on patient and family participation in end-of-life care has increased awareness of do-not-resuscitate (DNR) orders on the part of patients and health care professionals. Guidelines presently exist for the use of DNR orders in critically ill patients.1 These guidelines, which were initially published in 1991, clearly state that physicians should discuss patient preferences with regard to resuscitation efforts if the patient is at increased risk for cardiac or pulmonary failure. An increase in the use of DNR orders has been noted since the publication of these guidelines,2 particularly in elderly patients and those with comorbid illnesses.3- 6
Cardiovascular disease is the leading cause of death in the United States, accounting for almost 1 million deaths annually.7 Variations in the use of DNR orders in patients with cardiovascular disease appear to exist. A study of patients admitted with acute stroke noted that approximately 20% of patients had DNR orders written during their hospitalization.8 In contrast, the prevalence of DNR orders in patients with congestive heart failure was less than 5%.9 Data related to the use of DNR orders in patients with acute myocardial infarction (AMI) and trends in the use of these orders over time remain scarce.
The Worcester Heart Attack Study is an ongoing longitudinal investigation of residents of the Worcester, Mass, metropolitan area hospitalized with validated AMI.10- 12 This prospective study offers a unique opportunity to examine use of DNR orders in a community-wide sample of patients hospitalized with AMI. The purpose of the present study was to examine the use of DNR orders and associated patient characteristics in patients hospitalized with AMI betweem1991 and 1999. In addition, we examined the use of different treatment regimens and hospital outcomes in patients with DNR orders in this large population-based sample of patients with confirmed AMI.
This study is part of an ongoing population-based investigation that is examining changes over time in the incidence and case-fatality rates of residents hospitalized with a discharge diagnosis of AMI at all metropolitan Worcester hospitals. The details of this project have been described elsewhere.10- 12 In brief, the medical records of residents of the Worcester metropolitan area (1990 census estimate, 437 000 population) hospitalized for possible AMI were individually reviewed and validated according to predefined diagnostic criteria. These criteria consisted of a clinical history with findings suggestive of AMI, serum cardiac enzyme level elevations, and serial electrocardiographic findings consistent with evolving AMI.10- 12 At least 2 of these 3 criteria needed to be satisfied for study inclusion. All autopsy-proved cases of AMI were included irrespective of the primary diagnostic criteria. Cases of perioperative-associated AMI were not included. A total of 4621 residents of the Worcester metropolitan area hospitalized with validated AMI during 5 annual periods (1991, 1993, 1995, 1997, and 1999) constituted the population of this report.
Sociodemographic, medical history, and clinical data were abstracted from the hospital medical records of residents of the greater Worcester area with confirmed AMI. Information was obtained about the patient's age, sex, and prior comorbidities, including diabetes mellitus, hypertension, heart failure, angina, and stroke. Data regarding characteristics of the AMI were also collected, including AMI order (initial vs prior), type (Q wave vs non–Q wave), and location (anterior vs inferior/posterior). Information about the occurrence of clinically significant hospital complications, including heart failure and cardiogenic shock, was collected through the review of medical records in a standardized manner.13,14
Use of different treatment modalities, including invasive coronary procedures and beneficial cardiac medications (eg, aspirin, β-blockers, angiotensin-converting enzyme inhibitors, and thrombolytics) during the index hospitalization, was recorded from hospital medical records.15 Information about the use of DNR orders was collected through the review of hospital records and the physician's progress notes. Data on the timing of DNR orders were also collected from the medical chart, primarily during the latter study years.
Differences in the distribution of demographic, medical history, and clinical characteristics between AMI patients with and without DNR orders were examined through the use of χ2 and 2-sided t tests for discrete and continuous variables, respectively. Univariate analyses were performed to compare patients with DNR orders who survived to hospital discharge with those who died during the index hospitalization. The significance of changes during the nearly decade-long study period in the use of DNR orders was examined through the use of χ2 tests for trends. A logistic multivariable regression approach was used to examine the association between demographic characteristics, medical history, and clinical characteristics (predictor variables) and the use of DNR orders (outcome variable) during the acute hospitalization. Data for several comorbidities were available only for recent study years (including cancer, renal insufficiency, liver disease, and pulmonary disease) and thus were missing in a large proportion of the total cohort. Therefore, these variables were not included in the main regression models examining factors associated with DNR orders. We also examined the relation between use of DNR orders and receipt of cardiac medications and coronary interventions while controlling for potentially confounding demographic, medical history, and clinical variables (including development of heart failure and cardiogenic shock during hospitalization). A subgroup analysis was performed examining the relation between timing of DNR orders and hospital outcomes in patients in whom this information was available.
Approximately one fifth (19.4%) of the study sample had a DNR order noted in their medical records during the acute hospitalization. A significant increase in the use of DNR orders in patients with AMI was observed during the approximately decade-long study period (Figure 1). Utilization rates increased from 16% in 1991 to 25% in 1999 (P<.001).
Information about the timing of DNR orders was obtained in 542 patients (60% of patients with DNR orders). This information was primarily available from patients hospitalized in the 2 most recent study years (1997 and 1999). Among these patients, 55% had DNR orders noted in their medical records during the first day of hospitalization, 21% between 1 and 3 days, and 24% thereafter. Only 24 patients had DNR orders noted in their medical records before the current hospitalization as documented by admission records, including physician admitting notes. These patients were primarily elderly (age, ≥75 years).
Patients who had DNR orders were more likely to be older and female and have more comorbid illnesses (Table 1). Patients with DNR orders were also more likely to have multiple preexisting morbidities. The prevalences of 1, 2, or more comorbidities were 24%, 33%, and 31%, respectively, for patients with DNR orders compared with 34%, 25%, and 16% for patients without DNR orders. Patients with DNR orders were significantly more likely to have a prior and/or non–Q wave AMI compared with patients without DNR orders. Patients with DNR orders were significantly more likely to have development of heart failure and cardiogenic shock during hospitalization compared with patients who did not have a DNR order.
Differences in the use of beneficial cardiac therapies were observed between the respective comparison groups (Table 1). Patients with DNR orders were significantly less likely to have received effective cardiac medications during hospitalization for AMI, including aspirin, β-blockers, and thrombolytics. Digoxin was used in a greater proportion of patients in the DNR group. Patients with DNR orders were less likely to have undergone invasive procedures, including cardiac catheterization, percutaneous coronary intervention (PCI), and coronary artery bypass grafting, compared with patients without DNR orders.
As expected, patients with DNR orders experienced significantly higher hospital death rates compared with patients without DNR orders (44% vs 5%) (Table 1).
We examined possible changes over time in the demographic characteristics, clinical characteristics, and treatment practices of patients assigned DNR orders during the initial study year (1991), midpoint (1995), and most recent year under study (1999) (Table 2). Patients with DNR orders were consistently more likely to be elderly and female. For the comorbidities in which information was available for all 3 years, patients with a history of stroke, heart failure, and/or diabetes were more likely to receive DNR orders. For the most recent study year (1999) in which data were collected on a large number of comorbidities, patients receiving DNR orders were more likely to have a variety of comorbid illnesses present, including cancer, coronary disease, peripheral vascular disease, pulmonary disease, and renal insufficiency. Patients with prior and non–Q wave AMIs were also more likely to have DNR orders. Patients with DNR orders were more likely to die during hospitalization and to experience heart failure and cardiogenic shock during each of the years under study.
Increased use of effective cardiac medications, such as aspirin and β-blockers, was observed over time (Table 2). The use of these medications, however, remained significantly lower in patients with compared with those without DNR orders. Use of thrombolytics declined during the study period, and these agents were used less often in patients with DNR orders. Digoxin was used more often in patients with DNR orders.
Use of PCI increased over time, although differences remained in the use of these procedures according to DNR status. Patients with DNR orders were more likely to undergo cardiac catheterization and PCI in the latter half of the decade. However, these procedures were used consistently less often in patients with than in patients without a DNR order (Table 2).
Given the univariate associations observed between various demographic and clinical characteristics with the use of DNR orders, we more systematically examined these associations in a multivariable regression analysis (model 1) (Table 3). Advancing age, female sex, and histories of diabetes mellitus and stroke were significantly associated with the receipt of DNR orders. Patients assigned to DNR status were more likely to have experienced a prior and/or non–Q wave MI, and development of heart failure and death was more likely in these patients during the acute hospitalization. Study year was also associated with DNR status, reinforcing previously observed significant increases in the use of DNR orders over time.
A second multivariable adjusted regression model (model 2) was created to examine the use of specific cardiac medications in relation to DNR status, in addition to controlling for previously described demographic and clinical variables examined in the initial regression model. Similar results were seen as with the first model, with few exceptions. The occurrence of cardiogenic shock was associated with DNR status in the second regression model, and study year was more strongly associated with assignment of DNR orders.
Treatment with aspirin, β-blockers, thrombolytics, and cardiac catheterization was used significantly less often in patients with DNR orders after controlling for patient demographic characteristics, comorbidities, hospital complications, and length of hospital stay (Table 3). Use of PCI and coronary artery bypass grafting was also lower in patients with DNR orders, but did not reach statistical significance.
Since the timing of DNR orders may affect the receipt of beneficial cardiac medications and interventional procedures, we examined the association between baseline characteristics and cardiac therapies with timing of DNR orders in subjects in whom this information was available (Table 4). Patients with early institution of DNR orders (within the first 24 hours of hospitalization) were older and more likely to be women and to have a history of heart failure and/or cancer compared with patients who received DNR orders at a later time. Development of heart failure or cardiogenic shock was less likely in patients who received DNR orders early during their hospitalization, compared with patients who received DNR orders at a later time. When heart failure or cardiogenic shock did occur among patients with DNR orders, DNR orders often preceded these complications (Figure 2). Hospital case-fatality rates were higher in patients who received DNR orders more than 3 days into their hospital course (Table 4).
Since the timing of DNR orders may affect the receipt of cardiac medications, we examined timing of DNR orders in relation to the use of several medications and cardiac procedures. Use of most cardiac medications was similar according to timing of DNR orders (Table 4). The exceptions were patients who received DNR orders after 3 days of hospitalization who were more likely to receive calcium antagonists and digoxin. Patients who received DNR orders at a later time during their hospitalization were more likely to have undergone cardiac catheterization and PCI or coronary artery bypass grafting compared with patients who received DNR orders during the initial days of hospitalization. These interventions were more likely to occur before the DNR order was written (Figure 2).
To further explore the relation between receipt of cardiac therapies and use of DNR orders, we examined the use of medications at the time of hospital discharge for patients according to DNR status. Patients with DNR orders were significantly less likely to have received aspirin (odds ratio [OR], 0.65; 95% confidence interval [CI], 0.51-0.82), β-blockers (OR, 0.51; 95% CI, 0.41-0.64), and angiotensin-converting enzyme inhibitors (OR, 0.79; 95% CI, 0.63-0.99) compared with patients who did not receive DNR orders after controlling for age, sex, AMI severity, and major comorbidities. In contrast, at the time of hospital admission, patients with and without DNR orders had relatively similar rates of use of each of these cardiac medications.
The results of this community-wide study in a large sample of residents from a representative northeastern metropolitan area with confirmed AMI demonstrate a significant increase in the use of DNR orders from 1991 to 1999. As expected, patients assigned DNR orders were more likely to be older and female and include those with significant comorbid conditions. Patients with DNR orders were significantly more likely to die during hospitalization and to have development of clinically significant complications. Use of effective cardiac medications increased over the study period for patients with DNR orders. However, use of these cardiac medications was lower in patients with compared with those without DNR orders, irrespective of the timing of these orders. Similar findings were observed for the use of invasive coronary procedures, even after controlling for potentially confounding demographic and clinical factors.
The use of DNR orders in hospitalized patients has ranged widely (<1% to >75%), depending on patient- and disease-associated characteristics.3 Among patients with cardiovascular disease, data are extremely limited. These findings also vary widely depending on patient and provider characteristics.
In a large study of patients admitted with stroke to 30 hospitals in the Cleveland, Ohio, area between 1991 and 1994, 22% of patients received a DNR order during hospitalization.8 Similar prevalence rates of DNR orders in patients with acute stroke have been noted in other observational studies.3,16 Use of DNR orders in patients with acute cardiac disease has been lower than those observed for stroke patients.6 The prevalence of DNR orders in patients with congestive heart failure has been reported to be as low as 5%.17 In a recent examination of data from the Study to Understand Prognosis and Preferences for Outcomes and Risk of Treatments (SUPPORT) Project, nearly one quarter of patients with heart failure from 1989 to 1994 were assigned to DNR status.9 However, only one sixth of patients admitted with heart failure in this study had a written DNR order at the time of hospital discharge. We observed prevalence rates of DNR orders among patients admitted with AMI in the past decade similar to those observed in patients with heart failure in the SUPPORT study and to stroke patients hospitalized between 1991 and 1994.8,9 On the other hand, approximately 4% of patients admitted with AMI in a cohort of more than 14 000 Medicare patients hospitalized for a variety of illnesses had DNR orders.3 This population included patients hospitalized in the early to middle 1980s, which may partially explain the lower DNR rates observed compared with those noted in our more contemporary study.
Of interest, the use of DNR orders increased in our community-wide sample of patients hospitalized with AMI over time. A more than 50% increase in the use of DNR orders was observed for patients admitted with AMI in 1999 compared with patients admitted in 1991. This trend coincides with increased awareness regarding end-of-life issues and with the findings from other studies.2,18 In a small cross-sectional study, the use of DNR orders in patients dying of cardiovascular disease increased from 27% in 1983 to 64% in 1986.18 In a considerably larger study, data from nationally representative samples of Medicare patients also demonstrated an increase in the use of DNR orders over time (10% in 1981-1982; 13% in 1985-1986). This sample represented patients hospitalized with a variety of illnesses.2
Studies carried out in predominantly noncardiac patient populations have observed DNR orders to be more commonly used in the elderly,3,5,6,19- 21 women,3,19 and patients with other comorbid illnesses.6,20 In our study, older age and female sex were associated with receipt of DNR orders. The presence of comorbid conditions, including cancer, diabetes, heart failure, lung disease, peripheral vascular disease, and stroke, was also more prevalent among patients who had DNR orders during their hospitalization. Some evidence suggests that women and the elderly are less likely to want more aggressive treatment when acutely ill and thus may ask for DNR status more often.22 As expected, serious medical complications were more likely to develop in patients who received a DNR order during their hospitalizations.
Numerous studies have documented the high mortality in patients with DNR orders. Hospital death rates for patients with DNR orders have exceeded 50% in several studies.5,19,23 These studies include patients with a variety of diseases, including cancer and other noncardiovascular diagnoses. In our study, we found that approximately 44% of patients admitted with AMI who received DNR orders died during their hospitalization during the decade-long study period. Unfortunately, owing to our methods of data collection, we were unable to determine the proportion of hospital deaths that were attributed to withdrawal of life support. A small reduction in hospital case-fatality rates was observed in patients with DNR orders between 1991 (48%) and 1999 (38%). One possible explanation for the declining death rates in patients with DNR orders is that the overall treatment of patients with AMI has improved during the past decade; alternatively, DNR orders may be considered more often in less critically ill patients.
Evidence suggests that patients with DNR orders receive less aggressive management approaches.19,23 In contrast, other investigators have observed no change in the level of care delivered to patients before and after the receipt of DNR orders.20 We compared the use of several cardiac medications and coronary interventional approaches in patients according to DNR status and the timing of DNR orders. Patients with DNR orders often did not receive effective cardiac medications used in the management of AMI and were less likely to receive myocardial reperfusion strategies. Patients with DNR orders may also have been less likely to undergo coronary revascularization, since they might not qualify for cardiac catheterization at the institutions under study.
It is not clear whether patients who decide not to undergo cardiopulmonary resuscitation if their heart stops also decide not to receive these cardiac medications or interventions, although this remains a possibility. The decision to have comfort measures only may include the decision to withdraw cardiac-related medications. We did not have information available as to which DNR patients specified comfort measures only. This lack of information may partially explain some of the treatment differences observed between patients with and without DNR orders. Evidence suggests that physicians interpret DNR status differently in terms of the use of nonresuscitative measures, including medications.24 Furthermore, patients who obtained DNR orders late during their hospitalization also did not receive these medications at the same rates as patients who never had a written DNR order. The relative lack of receipt of effective cardiac medications and treatment approaches in patients with DNR orders are likely due to a variety of factors and influences. The reasons for these differences need to be more fully explored in future studies.
The strengths of this study include the large community-based sample of patients with confirmed AMI from all area hospitals and the ability to examine trends in the use of DNR orders throughout the 1990s. The study was carried out in men and women of all ages from a well-defined metropolitan area with demographic characteristics similar to US residents, enhancing the potential generalizability of our study findings. However, several potential limitations must be kept in mind in interpreting the results of the present study. Since this is an observational nonrandomized study, appropriate caveats need to be placed in the interpretation of the association of DNR status with patient characteristics and receipt of cardiac modalities. Unmeasured factors might have differed between those patients who received and those who did not receive DNR orders overall and during the different years under study. We also did not collect information about other factors, including nursing home status, history of dementia, or family members' wishes that may have affected the receipt of DNR orders. In addition, this cohort consists largely of white patients and thus may lack generalizability to other racial/ethnic groups. However, these findings serve as a starting point for further study and discussions about end-of-life issues in the patient population with acute and chronic cardiac illness.
Technological and treatment advances in the treatment of patients with AMI have increased dramatically during the last several decades. At the same time, patients and health care providers have come to the realization that prolongation of life, particularly in patients with a poor prognosis and limited quality of life, may not be desirable. The use of DNR orders has been promoted as a means for preventing futile resuscitative attempts. Patients and health care providers have become increasingly aware that end-of-life issues, including the use of DNR orders, are an important component of overall health care. Despite having frequent histories of cardiovascular disease before hospitalization, most patients have not had a discussion regarding end-of-life care before their hospitalization. Our findings reinforce the small percentage of patients who presented at greater Worcester area hospitals with a documented DNR status, even among patients with histories of significant comorbidities. Further studies are needed to explore the reasons for possible differences in the assignment of DNR orders in different patient groups and the use of various treatment approaches in patients with DNR orders. These issues take on further importance given expected increases in our elderly population during the next several decades.
Corresponding author: Elizabeth A. Jackson, MD, MPH, Department of Medicine, Division of Cardiovascular Medicine, University of Massachusetts Medical School, 55 Lake Ave N, Worcester, MA 01655 (e-mail: firstname.lastname@example.org).
Accepted for publication May 13, 2003.
This study was supported by grant RO1 HL35434 from the National Heart, Lung, and Blood Institute, Bethesda, Md, and the Meyers Primary Care Institute, Worcester, Mass.
This study was made possible through the cooperation of the administration, medical records, and cardiology departments of participating Worcester, Mass, metropolitan area hospitals.