|Home | About | Journals | Submit | Contact Us | Français|
Patients with advanced chronic disease are frequently offered medical and surgical interventions with potentially large trade-offs between benefits and burdens. Little is known about the frequency or outcomes of treatment refusal among these patients.
To assess the frequency of, reasons for, factors associated with, and outcomes of treatment refusal among older persons with advanced chronic disease.
Observational cohort study.
Two hundred twenty-six community-dwelling persons with advanced cancer, chronic obstructive pulmonary disease, or congestive heart failure, interviewed at least every 4 months for up to 2 years.
Participants were asked if they had refused any treatments recommended by their physicians, and why.
Thirty-six of 226 patients (16%) reported refusing 1 or more medical or surgical treatments recommended by their physician. The most frequently refused interventions were cardiac catheterization and surgery. The most common reason for refusal was fear of side effects (41%). Treatment refusal was more frequent among patients who wanted prognostic information (10% vs 2%, p=.02) or estimated their own longevity at 2 years or less (18% vs 5%, p=.02). There was an increased risk of mortality among refusers compared with non-refusers (HR 1.98, 95% CI 1.02–3.86).
Refusal of medical and surgical interventions other than medications is common among persons with advanced chronic disease, and is associated with a greater desire for, and understanding of, prognostic information.
Medication non-adherence among patients with chronic disease is common.1,2 More than 50% of Medicare beneficiaries with multiple chronic diseases report skipping or discontinuing medications within the past year,3 frequently citing adverse side effects or disagreement with their physicians about the need for medication as factors in their non-adherence.3,4 These reasons suggest that older persons with chronic diseases may weigh the benefits and burdens of therapy differently than their physicians do, or that their treatment goals may differ from the goals of their physicians.
Patients with advanced chronic disease face additional treatment options, including medical and surgical interventions, with even larger potential trade-offs between benefits and burdens. Frequent treatment refusal among these patients may be a marker for insufficient patient-centered decision making, as patients may be refusing treatments, which they perceive as failing to meet their treatment goals. However, little is known about the frequency or outcomes of non-adherence to, or refusal of, medical and surgical interventions among patients with advanced chronic disease.
The few studies that have looked at the refusal of treatments other than prescription medications have shown widely varying results. In 1 small retrospective study of older community-dwelling persons, nearly 40% of patients refused medical treatments or testing during the 6 months before their death.5 In 2 large studies of cardiac catheterization after myocardial infarction, refusals occurred in under 3% of cases and were not associated with decreased survival at 1 year.6,7 And in 1 study of women with newly diagnosed breast cancer, refusal of surgery occurred in under 2% of cases, and refusal was associated with decreased survival at 5 years.8 To date, there have been no studies looking at the refusal of a range of medical and surgical interventions among patients with advanced chronic disease.
We set out to describe, in a cohort of community-dwelling persons with advanced chronic disease, the frequency of refusal of treatments for these diseases, the reasons given by patients for their refusal, patient factors associated with refusal, and the effect of refusal on mortality.
Participants in this study were the members of a cohort study designed to examine the treatment preferences of older persons with advanced illness.9 Two hundred twenty-six community dwelling persons 60 years of age or older with advanced illness secondary to cancer, congestive heart failure (CHF), or chronic obstructive pulmonary disease (COPD) were recruited from oncology, cardiology, or pulmonary practices in 3 metropolitan areas in southern Connecticut, the ambulatory clinics of 2 Veterans Affairs (VA) hospitals, and among inpatients from 3 hospitals (one each university, Veterans, and community).10 Enrollment was stratified by diagnosis. Advanced illness was defined according to the clinical criteria used by the Connecticut Hospice11 or the criteria used in the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT) (Appendix 1)12,13 and the need for assistance with at least 1 instrumental activity of daily living (IADL).14,15Given this need for assistance, all subjects had an informal caregiver and were asked to provide the name of a proxy contact who was used to confirm patient outcomes if the patient could not be reached. Patients were excluded if they had moderate to severe cognitive impairment as measured by the Short Portable Mental Status Questionnaire16 and the EXIT, a test of executive functioning,17 or if they were not full-time residents of Connecticut. The study protocol was approved by the human investigations committee of each of the participating hospitals. All patients provided written informed consent.
Of the 548 patients initially identified by chart review, 30 had physicians who refused permission for patient contact, 24 died before telephone screening, 19 declined screening, and 6 could not be reached. Of the 469 patients screened, 108 were excluded because they required no IADL assistance, 76 because of cognitive impairment, and 6 because they were not full-time Connecticut residents. Of the 279 eligible patients, 51 refused participation and 2 died before participation, resulting in 226 participants. Non-participants did not differ from participants according to age or sex. Of eligible patients with CHF, 8.3% refused participation, compared with 20.2% of patients with cancer and 25.0% of patients with COPD (P=.02).
All participants were interviewed in their homes by trained research staff at least every 4 months for up to 2 years. In addition, participants were contacted by telephone every month and asked about changes in health status, defined as either: 1) a new disability in a basic ADL (BADL),18 2) a prolonged hospitalization (≥7 days) or one resulting in discharge to a nursing home, sub-acute facility, or rehabilitation facility, or 3) introduction of hospice services. Changes in health status triggered an additional interview immediately. Patient deaths were also ascertained through proxy informants during these monthly calls.
At their initial interview, patients were asked, “Have there been any treatments that your doctor recommended for you that you decided not to have?” At all follow-up interviews patients were asked, “Has there been any treatment that your doctor recommended that you decided not to have since the last interview?” Data generated from the initial interviews is referred to as retrospective, and from the follow-up interviews as prospective. We separated these cohorts because they represent 2 distinct groups of patients. The retrospective cohort includes patients who refused an intervention and survived long enough in a community setting to be eligible for the parent study, whereas the prospective cohort includes all patients who refused a treatment regardless of their outcome. The prospective cohort included 189 subjects because, of the 226 participants, 8 withdrew after the initial interview (4%), 26 died before completing a follow-up interview (12%), and 3 were too ill to participate in follow-up interviews (1%).
Patients were asked, “Was the treatment you decided not to have: hospitalization, surgery, dialysis, cardiac catheterization, intubation, chemotherapy, radiation therapy, transplantation, or other [procedure]?” These treatments were chosen because they are frequently offered and pose significant trade-offs between their potential benefits and burdens to the individual patient. At each interview, patients were also asked if they had undergone any of these treatments (except transplantation).
For each treatment refused, patients were offered a list of possible reasons for refusal. They were asked, “Did you a) think it would not work, b) think that the side-effects would be too much, c) think you would do better with a different treatment, d) not want to do anything to prolong your life, or e) have some other reason?” Patients could choose multiple reasons for each treatment refused.
To describe the population and identify factors associated with treatment refusal, data on patients’ sociodemographic, health, and psychosocial status, as well as experience with health care decision making were collected at the initial interview. These variables were selected based on their association in prior studies with decisions to forgo cardiopulmonary resuscitation,19,20 and clinical experience. Sociodemographic variables included age, sex, race, marital status, and education. Health status variables included functional status measured using the BADL scale.18 Psychosocial factors included self-rated longevity, measured by asking patients, “If you had to guess, how long do you think that you might have to live?”, desire for prognostic information, measured by asking, “Would you want your doctor to say how long you have to live?”, and quality of life measured by asking patients to rate their current health on a scale from poor to excellent. In addition, patients were asked about the presence of a living will, their knowledge of alternatives to hospitalization should their disease progress, and whether they had ever personally cared for a chronically ill loved one.
To describe the overall frequency of treatment refusal, we first determined the proportion of participants who refused 1 or more treatments in both the retrospective and prospective cohorts. Using the treatment refused as the unit of analysis, we then calculated the rate of refusal by dividing the number of refusals by the number of times the treatment was recommended, defined as the number of times it was undergone plus the number of times it was refused. To examine factors associated with refusal, the participant was the unit of analysis, and we used the Fisher exact test.21 We used survival analysis to examine the association between treatment refusal and mortality in the prospective cohort. Specifically, treatment refusal was treated as a time-dependent variable in a Cox regression model,22 and the model was adjusted for diagnosis as well as factors that were associated with treatment refusal in the bivariate analysis described above. SAS version 9.1.3 (SAS Institute, Cary, NC) was used for all analyses and a p value less than 0.05 (two-sided) was used to indicate statistical significance.
Patient characteristics are summarized in Table 1. The large percentage of male participants reflects patient recruitment from both VA and non-VA settings.
Overall, 36 (16%) of 226 patients reported refusing 1 or more medical or surgical treatments recommended by their physician. Twenty-nine people refused 1 treatment, 6 refused 2 treatments, and 1 person refused 4 treatments. Thus, a total of 45 individual treatments were refused (Table 2). The pattern of refusals differed between the retrospective and prospective cohorts. In the retrospective cohort, only surgery and cardiac catheterization were refused with any significant frequency (11% and 13%, respectively). In contrast, the refusal rate for each intervention in the prospective cohort, with the exception of hospitalization, was over 5%.
As shown in Table 3, the most frequently chosen reason for treatment refusal was concern over side effects (41%). Patients also frequently chose “other reason” and gave additional answers that were varied and highly individual. This is illustrated by the following quotations from patients who refused cardiac catheterization, transplantation, and hospitalization, respectively:
Wanted to delay it. Was concerned about my wife at home and wanted to get back to her. I preferred to wait and see.
I’m too old. Young people should get the hearts that are available.
They probably would not let me come home again.
In the retrospective cohort, 17% of patients with CHF refused treatments, compared with 3% of patients with cancer or COPD (p=.0007, Table 4). Refusal was also more frequent among patients who wanted their physician to tell them how long they had left to live (10% vs 2%, p=.02), or estimated their own longevity as 2 years or less (18% vs 5%, p=.02). Patients who lived alone refused treatments more often (13% vs 5%), as did those with prior experience caring for a chronically ill loved one (9% vs 4%), but these differences did not reach statistical significance. Neither sociodemographic status (age, sex, race, marital status, level of education, or income), health or functional status, the presence of a living will, nor knowledge of alternatives to hospitalization were associated with treatment refusal, and there was no difference in functional status or self-reported quality of life between refusers and non-refusers. In the prospective cohort there were no statistically significant differences between refusers and non-refusers.
The mortality rate over 2 years of follow-up in the prospective cohort for patients with cancer, COPD, and CHF, was 42%, 29%, and 27%, respectively. Treatment refusal in this cohort was associated with an increased risk of mortality in unadjusted analysis (hazard ratio [HR] 2.80, 95% confidence interval [CI] 1.56–5.05). The effect was attenuated after adjustment for diagnosis, self-rated longevity, the desire for prognostic information, living alone, and having previously cared for a seriously ill loved one (HR 1.98, 95% CI 1.02–3.86).
Among a cohort of community-dwelling older persons with advanced chronic disease, refusal of medical and surgical interventions other than medications was reported with clinically significant frequency. Fear of side effects was the most common reason cited for refusal. Many patients also provided other reasons that were highly personal. Reports of treatment refusal were more common among patients with CHF compared to those with cancer or COPD, and among patients with both a greater desire for, and understanding of, prognostic information compared to those with less desire for, and understanding of, prognostic information. Treatment refusal was associated with an increased risk of mortality, which persisted after controlling for diagnosis and other potential confounders.
Our finding that 16% of persons with advanced chronic disease refuse medical and surgical interventions recommended by their physicians lies between previous estimates of treatment refusal in the literature. Refusal rates for cardiac catheterization by patients with myocardial infarction were between 2% and 3%,6,7 and in 1 study of women with newly diagnosed breast cancer, 1.3% refused surgery.8 In contrast to these low refusal rates for single interventions, up to 40% of outpatients in the community were shown to refuse diagnostic or therapeutic procedures offered to them in their last 6 months of life.5 This increase in treatment refusals, from under 3% in the setting of a new diagnosis to nearly 40% at or near the end of life, suggests that as illness severity increases so does a patient’s willingness to forgo potentially burdensome interventions. This notion is supported by the frequency of refusal in our patient population, whose illness severity lies intermediate between those of the prior studies, and by our finding that refusal rates increased from the retrospective cohort to the prospective cohort, the latter including patients with more advanced illness.
The frequently cited reasons for treatment refusal in this study expands on prior work showing that fear of side effects plays a major role in treatment decision making for prescription medications. Over 17% of Medicare recipients reported skipping doses or stopping medications outright because of side effects,3 and more than 15% of patients with lupus nephritis stated they would prefer a less efficacious medical regimen to avoid the toxicity associated with cyclophosphamide.23 Such findings support the notion that patients are strongly influenced by their perception of side effects, regardless of whether a treatment is indicated from the physicians’ point of view. The personal concerns cited as reasons for refusal also confirms prior work showing that treatment decision making is influenced by a diverse range of personal experience.24
Prior investigation identified an association between knowledge of limited prognosis and preferences for care aimed at comfort rather than life-prolongation.25 The findings of the current study expand upon this work to show that prognostic understanding is not only associated with preferences, but also with actual decisions to forgo interventions. Moreover, in addition to understanding of prognosis, desire for prognostic information was associated with treatment decision making, suggesting that patients who are more accepting of their disease and its prognosis may have treatment goals that differ from those of other patients.
There are 2 possible interpretations of the finding of an increased mortality risk associated with treatment refusal. One is that treatment refusal may hasten death by forgoing life-prolonging treatments. Unfortunately, we did not have the detailed clinical data that would be necessary to control for prognostic differences between patients who refused and those who did not. However, because all patients in the study met the criteria for advanced illness, it is not clear that further prognostic accuracy would be possible. The other possibility is that this finding reflects an increased frequency of refusal among patients with more advanced illness, an interpretation that is supported by our finding that refusal rates increased from the retrospective cohort to the prospective cohort.
This study had several important limitations. First, as a secondary data analysis the study was not powered specifically to examine the outcome of treatment refusal. Second, the large proportions of male and white patients may limit generalizability. Third, by relying solely on patients’ reports of recommended and refused treatments, the possibility of recall bias exists. Fourth, we do not know what treatment alternatives, if any, were offered by physicians after a refusal had occurred. It is possible that the treatment refusal represented the first step in a process to elicit patients’ values regarding the benefits and burdens of interventions, and was followed by discussion of an alternative intervention. It is unlikely, however, that patients were offered additional options after their refusal, because prior work has shown that patients are rarely provided with treatment alternatives by their physicians or surgeons.26 Moreover, even if such a process did occur, the presence of a treatment refusal suggests that alternatives were offered sequentially rather than simultaneously. This approach has the disadvantage of requiring a patient to take the active step of turning down a treatment recommended by his or her physician, and it is unlikely that many patients feel empowered to do this.27 Further work is needed to examine physician–patient behaviors and communications regarding these decisions. Fifth, assessment of treatment refusals was done using a single-item question, which has not undergone testing for validity or reliability. Finally, our results describing the factors associated with refusal in the retrospective cohort were not replicated in the prospective one, suggesting that more work is needed to elucidate the relationship between patient’s understanding of their disease and treatment refusal.
Our study highlights the frequency with which older persons with advanced illness refuse treatments recommended by their physicians. This frequency of refusal suggests that physicians may be recommending treatments to these patients that pose unacceptable burdens or that fail to meet patients’ goals. These patients may require a broader range of treatment alternatives, so that they can select the option that best meets their goals of care.
Supported by grant PCC-98-070-1 from VA HSR&D, R01 G19769 from the National Institute on Aging, P30 AG21342 from the Claude D. Pepper Older Americans Independence Center at Yale, and a Paul Beeson Physician Faculty Scholars Award. Dr. Rothman is a postdoctoral research fellow in geriatics and clinical epidemiology and is supported by a training grant from the National Institute on Aging (T32AG1934). Dr. Fried is supported by K02 AG20113 from the National Institute on Aging. The authors thank Carm Joncas, RN and Barbara Mendes, RN for their extraordinary interviewing skills. Portions of this work were presented previously at the 2005 Annual Meeting of the American Geriatrics Society.
Conflict of Interest Statement None disclosed.
Chronic Obstructive Pulmonary Disease:
Congestive Heart Failure: