Groenewoud JH, van der Heide A, Kester JGC, de Graaff CLM, van der Wal G, van der Maas PJ. A Nationwide Study of Decisions to Forego Life-Prolonging Treatment in Dutch Medical Practice. Arch Intern Med. 2000;160(3):357-363. doi:10.1001/archinte.160.3.357
Decisions to withhold or withdraw life-prolonging treatment in terminally ill patients are common in some areas of medical practice. Information about the frequency and background of these decisions is generally limited to specific clinical settings. This article describes the practice of withholding or withdrawing life-prolonging treatment in the Netherlands.
Questionnaires were sent to the attending physicians of a stratified sample of 6060 of all 43,002 cases of death in the Netherlands from August 1 through November 30, 1995. The questions concerned the treatments foregone, the patient characteristics, and the decision-making process. The response rate was 77%.
A nontreatment decision was made in 30% (95% confidence interval, 28%-31%) of all deaths in the Netherlands in 1995; this is an increase compared with 28% (95% confidence interval, 26%-29%) in 1990; in 20% of all deaths, this decision was the most important end-of-life decision. Artificial nutrition or hydration and antibiotics were the treatments most frequently foregone, each accounting for 25% of cases in which a nontreatment decision was made. Nursing-home physicians withheld or withdrew treatment more often than clinical specialists or general practitioners in 52%, 35%, and 17% of all deaths they were involved with, respectively. Of the patients in whom a nontreatment decision was the most important end-of-life decision, 26% were competent; of those, 93% were involved in the decision making. In 17% of patients, the nontreatment decision was made without being discussed with the patient or the patient's relatives and without knowledge of the patient's wishes. Life was shortened by an estimated 24 hours or less in 42% and 1 month or more in 8% of patients.
Decisions to forego life-prolonging treatment are frequently made end-of-life decisions in the Netherlands and may be increasing. Most of these decisions do not involve high-technology treatments, and the consequences, in terms of shortening of life, are relatively small.
ADVANCES IN medicine and technology have had a great impact on the care of critically ill patients. Life-sustaining interventions, like mechanical ventilation and artificial nutrition or hydration, enable physicians to effectively postpone death in many patients who otherwise would have died. There is a growing interest in a humane approach to dying, and the inherent question is whether use of life-sustaining therapy is always in the patient's best interest. Currently, the right of competent patients to refuse treatment is widely accepted, even if this should hasten death.
During the last decade, several empirical studies1- 7 were published on the attitudes of patients and physicians toward the limitation of life-prolonging treatment and on the occurrence of decisions to forego life-prolonging treatment in medical practice.
The present study was part of a nationwide study8,9 conducted in 1995-1996 in the Netherlands on medical end-of-life decisions, including decisions to withhold or withdraw potentially life-prolonging treatment (nontreatment decisions), that was partly a replication of a study done in 1990.10,11 The goals of the present study were to obtain reliable estimates of the incidence of nontreatment decisions; describe the treatments foregone; describe the characteristics of patients, physicians, and circumstances involved; explore the role of patients, their relatives, and medical staff in the decision making; and compare the results with those of the 1990 study.
Details of the data collection have been published previously.8 In summary, the procedure was as follows. All causes of death of all Dutch inhabitants are routinely reported to Statistics Netherlands, Voorburg, together with the age and sex of the deceased and the name of the reporting physician. From August 1 through November 30, 1995, 43,002 death certificates were received. For reasons of statistical efficiency, 2 physicians classified all cases of death in the age group of 1 year and older into 5 strata on the basis of increasing probability that a medical decision concerning the end of life was made: when the cause of death was not suspected of involving any medical decision concerning the end of life (eg, car accident resulting in sudden death), the death was assigned to stratum 0, and when the death certificate provided explicit indications about an end-of-life decision being made the death was assigned to stratum 4. A sample was drawn from each stratum: 112 of the deaths in strata 0 and 1,
18 in stratum 2,
14 in stratum 3, and
12 in stratum 4. Samples taken from stratum 0 were retained in the study, but no questionnaires were sent. All deaths within the first year of life were included in the study. For these deaths and all deaths in strata 1 to 4, the reporting physician received a written questionnaire with 24 questions. Nonresponding physicians received a reminder 3 weeks after the questionnaire was sent. Of 6060 questionnaires that were mailed, 77% were returned and nearly all contained complete information. The results of this study are based on 5146 cases of death. Anonymity was guaranteed by interposition of a notary who removed the sample serial numbers from the returned questionnaires. The 1995 study covered the same period of the year as the earlier study in 1990.
The physician was asked whether any medical decision concerning the end of life had preceded death, namely decisions to withhold or withdraw potentially life-prolonging treatment, to intensify alleviation of pain and symptoms with possible life-shortening effects, and to administer or provide drugs explicitly intended to hasten death. Questions were asked whether the end-of-life decision had been made taking into account the possibility that death would be hastened or whether the physician had had the partial or explicit intention of hastening death to assess the moral and clinical weight of the decision. When more than one end-of-life decision had been made, the most important decision was defined as the weightiest decision, while the administration of drugs was considered to be more decisive than the withholding or withdrawal of treatment in case of equal weights. With respect to the most important end-of-life decision, questions were included about the decision-making process and the patient's competence. The questionnaire was very similar to the 1990 questionnaire12 except for an additional question about the type(s) of treatment foregone in the case of a nontreatment decision.
Death was defined as nonsudden if the respondent answered no to the question "Was death sudden and totally unexpected?"
A decision to withhold or withdraw potentially life-prolonging treatment was classified as the most important end-of-life decision when it was the only end-of-life decision made or when treatment was foregone with the explicit intention to hasten death in combination with a decision to alleviate pain and symptoms without the explicit intention to hasten death.
The patient's competence was described as the patient's ability to assess his or her situation and make an adequate decision.
In the Netherlands, general practitioners are family doctors who often have a long-standing relationship with their patients. They function as the gatekeepers of clinical medicine. Clinical specialists mainly work in hospitals, where they provide outpatient and inpatient care. Nursing-home physicians care predominantly for elderly patients with chronic diseases, physical or mental disorders, or disabilities who live in institutions. General practitioners and nursing-home physicians have 3 years of additional training after their basic training as medical doctors; clinical specialists have 5 to 6 years of additional training.
Weighted percentages valid for the whole population were calculated. Weights are based on the unequal sampling fractions in the strata, on corrections for nonresponse, and on the total number of deaths according to age and sex in 1995. This weighting procedure is described in detail elsewhere.12
In 1995, a decision to withhold or withdraw (possible) life-prolonging treatment preceded 30% (95% confidence interval [CI], 28%-31%) of all deaths in the Netherlands (Table 1) and 43% (95% CI, 42%-45%) of all nonsudden deaths. These figures show an increase compared with 28% (95% CI, 26%-29%) of all deaths and 39% (95% CI, 38%-41%) of all nonsudden deaths in 1990.11 Nontreatment decisions were made relatively frequently in patients aged 80 years and older (36%), in females (34%), and in deaths due to mental disorders (including Alzheimer dementia) (52%), disease of the digestive tract (50%), neurologic disease (including cerebrovascular accidents) (43%), or pulmonary disease (41%). In nonsudden deaths, logistic regression analysis, with a nontreatment decision being made or not as a dependent variable and age (3 categories), sex, and cause of death (7 categories) as independent variables, showed that age and cause of death, but not sex, independently contributed to the probability that a nontreatment decision was made. Table 1 also shows that nursing-home physicians made nontreatment decisions relatively often (52%) compared with other physicians (17% and 35% for general practitioners and clinical specialists, respectively). When the physician specialty was included in the logistic regression model, patient's age and cause of death and the physician specialty were associated independently with nontreatment decisions.
Of the nontreatment decisions, 49% involved only withholding life-prolonging treatment, whereas 51% involved withdrawal of life-prolonging treatment with or without withholding such treatment. The nontreatment decision was the most important end-of-life decision in 20% of all deaths. In another 8%, the decision to alleviate pain or symptoms was the most important end-of-life decision, and in 2% the physician performed euthanasia or assisted suicide after treatment had been foregone.
Of the respondents who reported a nontreatment decision, 86% specified the type(s) of treatment foregone: 60% mentioned 1 type of treatment foregone and 26% mentioned 2 or more different types of treatment foregone (a mean of 1.36 types of treatment per patient). As the wide variety of combinations of types of treatment could not easily be classified, results will be presented for all types of treatment separately.
Artificial nutrition or hydration and antibiotics were the treatments most frequently foregone (Table 2), each accounting for 25% of cases in which a nontreatment decision was made. Decisions to forego artificial nutrition or hydration in patients involved nutrition in 60%, hydration in 14%, and both nutrition and hydration in 26%. These decisions were predominantly made by nursing-home physicians: 55% of cases in which artificial nutrition or hydration was foregone (data not presented in Table 2). Vasopressors were foregone in 11% of cases, relatively frequently by clinical specialists (18%), and medications other than antibiotics or vasopressors were foregone in 18% of cases. Clinical specialists relatively frequently reported foregoing mechanical ventilation (22%) and surgery (14%), and general practitioners forewent chemotherapy or radiotherapy (9%).
Table 3 shows that decisions to forego artificial nutrition or hydration mainly concerned patients aged 80 years and older (68%), female patients (70%), and patients who died from cancer (24%), neurologic disease (24%), or a mental disorder (17%). Cancer was the most frequent cause of death in patients for whom artificial nutrition or hydration was foregone by general practioners (41% of patients) or clinical specialists (42% of patients), whereas, if this decision was made by nursing-home physicians, the most frequent cause of death was a mental disorder (mainly dementia) (26% of patients).
Compared with other types of treatment, mechanical ventilation and chemotherapy or radiotherapy were more often foregone in patients younger than 65 (32% and 43%, respectively). Antibiotics were mainly foregone in patients dying from neurologic disease (25%), pulmonary disease (19%), or cancer (18%). As might be expected, resuscitation (35%) and vasopressors (44%) were predominantly foregone in patients who died from cardiovascular disease, and chemotherapy or radiotherapy (94%), in patients with cancer.
Of the patients for whom the nontreatment decision was the most important end-of-life decision, 26% were competent, 67% were not fully competent, and in 7%, no information on the patient's competence was available. Nontreatment decisions made by nursing-home physicians more frequently concerned patients who were not fully competent (83%) than decisions made by clinical specialists (69%) or general practitioners (42%). Nontreatment decisions were discussed with 93% of competent patients (Table 4); in 67% of the competent patients, the decision was made at the patient's explicit request. There was no major difference between the treatments in discussing this decision; only the decision not to resuscitate had been discussed with the patient less often. The most frequently mentioned reasons for not discussing the decision with a competent patient were that "the decision was clearly the best one for the patient" (38%) and "discussion would have done more harm than good" (13%). Of patients who were not fully competent, 14% were involved in the decision-making process, and in another 13% of patients the physician had information about the patient's previous wish. The most frequently mentioned reasons for not discussing the decision with a patient who was not fully competent were that the patient was unconscious (51%) or that the patient had dementia (41%).
The nontreatment decision was discussed with colleagues relatively often when it concerned mechanical ventilation (92%), resuscitation (85%), or dialysis (76%) (Table 4). The patient's relatives were relatively frequently involved in the decision making when artificial nutrition or hydration (83%) or antibiotics (81%) were foregone. Of all nontreatment decisions, 17% were made without discussion with the patient or the patient's relatives and with no knowledge of the patient's wishes. In 4% of patients, the physician had discussed the decision with no one.
The physician estimated that life was shortened by 24 hours or less in 42% of nontreatment decisions, by 1 day to 1 week in 28%, by 1 week to 1 month in 15%, and by 1 month or more in 8% (no data were available on the estimated shortening of life in 7%). When life was shortened by 24 hours or less, relatively frequently, the decision had been made without involvement of the patient or the patient's relatives and without information about the patient's previous wishes (25%).
In the Netherlands, decisions to forego life-prolonging treatment are frequently made end-of-life decisions.8 The present study shows that the incidence of nontreatment decisions increased between 1990 and 1995,11 which may have resulted from increasing possibilities to sustain life in an ageing society together with a growing awareness among patients and physicians that preservation of life should not be the only goal of medicine.
Although several empirical studies on nontreatment decisions preceding death have been done elsewhere, to our knowledge, there is no other nationwide study. Most other studies concern a particular hospital or a particular department, mainly intensive care units,1,4,6,7,14 or a particular group of patients, such as patients with end-stage chronic diseases15 or elderly people.3 Therefore, comparability is limited.
In our study, artificial nutrition or hydration and antibiotics were the treatments most frequently withheld or withdrawn. Our results apparently differ in this respect from the many other studies on nontreatment decisions in which high-technology treatments were most often foregone. For instance, in intensive care units mechanical ventilation is the form of life support most frequently withdrawn.1,4,6 Two other studies2,5 on nontreatment decisions preceding hospital deaths, 1 of which involved only patients receiving acute care, showed that patients for whom dialysis, intubation, or mechanical ventilation was considered declined those particular treatments more frequently than if antibiotics or intravenous fluids were involved. Note that all these studies focus on particular patient categories for whom a particular treatment was considered or patients who already received that treatment. On the other hand, the pattern in these other studies corresponds to the results of a survey of physicians' attitudes toward the withdrawal of different forms of life-sustaining treatment under varying clinical circumstances, showing that physicians prefer to withdraw forms of life support that can be described as scarce, expensive, artificial, or high technology; of 8 forms of life support, blood products are the most likely to be withdrawn and intravenous fluids the least.16 Some studies2,4,5 conducted in the hospital or intensive care unit also mention a high proportion of do-not-resuscitate orders. In the 1990 study,10 unlike the 1995 study, the decision not to resuscitate in cases of cardiac or respiratory arrest was specifically studied: we found that do-not-resuscitate orders were in effect in a large proportion of hospital deaths (61%).17 In the present study, we did not ask respondents about anticipatory do-not-resuscitate decisions but about actual decisions to withhold or withdraw resuscitation that (probably or certainly) resulted in shortening of life.
Logistic regression analysis adjusting for the patient's age, sex, and cause of death showed a difference in nontreatment decisions between 3 major types of physician specialty; such decisions were made significantly more frequently by nursing-home physicians (52%) and clinical specialists (35%) compared with general practitioners (17%). Clinical circumstances and patient characteristics other than age, sex, and cause of death may account for these differences. The motives for making nontreatment decisions did not come within the scope of the present study, but from the 1990 study we know that the wish of the patient was the most important reason (74%) when the nontreatment decision was made at the patient's explicit request. For nontreatment decisions that were not made at the patient's explicit request, the most important considerations to forego treatment were "no chance of improvement" and "the futility of further treatment" (72%).11
Compared with euthanasia or physician-assisted suicide, nontreatment decisions have less far-reaching consequences in terms of shortening (or not prolonging) life. In euthanasia and physician-assisted suicide, the estimated shortening of life was 24 hours or less in 17%, 1 day to 1 week in 42%, 1 week to 1 month in 32%, and 1 month or more in 9%8; in nontreatment decisions, these percentages were 42%, 28%, 15%, and 8%, respectively (the shortening of life was unknown in 0% of euthanasia and physician-assisted suicide and 7% of nontreatment decisions).
Of the patients for whom a nontreatment decision was the most important end-of-life decision, 67% were not fully competent. Information about a patient's previous wish was available in 13% of not fully competant patients for whom a nontreatment decision was made (see "Results" section). Although there is a broad-based consensus that in such cases the patient's partner or relatives should be involved in the decision making, a number of nontreatment decisions in our study were made without discussion with the patient or his or her relatives. This lack of discussion is no exceptional finding. Of decisions to withhold cardiopulmonary resuscitation in incompetent patients admitted to a public teaching hospital, 21% had not been discussed with family members.18 Of physicians in intensive care units, 25% reported foregoing life-sustaining treatment because of futility of continuing treatment without the patient's or the patient's relatives' consent.14 Motives can also be practical; for example, relatives could not be found in one case.1 However, in exceptional cases, physicians in our study failed to discuss the nontreatment decision even with competent patients because they thought the decision was clearly the best for the patient.
Our study shows that decisions to forego life-prolonging treatment are frequently made end-of-life decisions and may be increasing. The future importance of these decisions is likely to be even greater, warranting more attention for these decisions in research, in the ethical debate, and in education of physicians and the general public. The quality of the decision making may benefit from openly studying and deliberating the subject, which especially holds for nontreatment decisions that for various reasons cannot be discussed with the patient or the patient's relatives.
Accepted for publication May 3, 1999.
The study was funded by a grant from the Ministry of Health, Welfare and Sports and the Ministry of Justice, the Hague, the Netherlands.
We thank Magda M. Henke-Kulakowska, MD, Henk Noort, MA, and Marjolein Smit, MD, for their contributions to the study; Karen Gribling-Laird, MA, for her grammatical advice; and Caspar W. N. Looman, MSc, and Gerard J. J. M. Borsboom, MSc, for their statistical advice.
Corresponding author: Johanna H. Groenewoud, MD, Department of Public Health, Erasmus University Rotterdam, PO Box 1738, 3000 DR, Rotterdam, the Netherlands (e-mail: firstname.lastname@example.org).