J Med Ethics 37:327-332 doi:10.1136/jme.2010.039248
  • Clinical ethics
  • Paper

Limitation of treatment at the end of life: an empirical-ethical analysis regarding the practices of physician members of the German Society for Palliative Medicine

  1. Jochen Vollmann1
  1. 1Institute for Medical Ethics and History of Medicine, Ruhr-University Bochum, Bochum, Germany
  2. 2German Society for Palliative Medicine; University Witten/Herdecke, Germany
  1. Correspondence to Dr Jan Schildmann, Institut für Medizinische Ethik und Geschichte der Medizin and History of Medicine, Ruhr-Universitaet Bochum, Malakowturm-Markstraße 258a, D-44799 Bochum, Germany; jan.schildmann{at}
  1. Contributors All authors contributed to the conception, design, analysis and interpretation of the data. All were involved in the production of the drafts of the article and have given their approval of the submitted version to be published. There is no one else who fulfils the criteria of authorship.

  • Received 25 July 2010
  • Revised 1 November 2010
  • Accepted 10 December 2010
  • Published Online First 2 February 2011


Objectives To determine the frequencies and types of limitation of medical treatment performed by physician members of the German Society for Palliative Medicine and to analyse the findings with respect to clinical and ethical aspects of end-of-life practices.

Design Cross-sectional postal survey.

Setting Data collection via the secretary of the German Society for Palliative Medicine using the German language version of the EURELD survey instrument.

Subjects All 1645 physician members of the German Society for Palliative Medicine.

Main outcome measures Types and frequencies of limitation of treatment and possible determinants.

Results 901 physicians participated in the study (response rate 55.8%). Participants reported limitation of treatment in 69.1% of cases. These decisions most often affected artificial nutrition (19%), chemotherapy (14%), antibiotics (11%) and medication other than antibiotics (11%). In the majority of eligible cases, physicians estimated the life-shortening effect of limitation of treatment to be <7 days. However, estimations differ depending on the medical measures in question. Bivariate statistical analysis indicated that withholding of treatment was performed significantly more frequently for patients aged ≥65 years (p=0.019). In addition, there were significant associations between the incidence of limitation of treatment and the different diseases reported by respondents as the underlying cause of death.

Conclusion The findings of this study provide information on the current state of an ethically and clinically challenging aspect of clinical practice and can serve as a starting point for further interdisciplinary research on normative and empirical aspects of treatment decision-making at the end of life.


Treatment decisions at the end of life pose ethical and clinical challenges. On the one hand there is an ever increasing number of treatment options available which may be administered also during the last phase of life. On the other hand, there is often limited evidence as to whether or not such treatment is of benefit to the patient. With this background it is not surprising that decisions about the limitation of treatment at the end of life are common and, at the same time, perceived by physicians as difficult tasks.1–3

In Germany as in many other countries there is an ongoing debate about an appropriate normative and practical framework for end-of-life practices in modern medicine. One characteristic of the German debate is the historical context. The murdering of patients under the label of ‘euthanasia’ during the Nazi regime has influenced the public and scientific debate and, for a long time, there has been a taboo with respect to an open discussion about physicians' end-of-life practices in medicine.4 5 A second characteristic of this debate—which is relevant not only for Germany but has also been discussed with respect to physicians' end-of-life practices in other countries6 7—is the frequent confusion regarding the distinction between acts and omission at the end of life and their relevance for the ethical and legal analysis. Part of this confusion in the German debate is routed in the terminology of ‘aktive Sterbehilfe’ versus ‘passive Sterbehilfe’. While the first term refers to the ending of a patient's life in the sense of the ‘active’ application of a substance (by the physician) without which the patient would not have died, the latter term refers to the limitation of medical treatment in the sense of the ‘passive’, letting the dying process take its course. However, in clinical practice as well as in the public debate, the use of the attributes ‘active’ and ‘passive’ are often understood with reference to the mode of action by the physician. Many physicians, for example, perceive ‘switching off’ a ventilator as (illegal) ‘aktive Sterbehilfe’ because the physician ‘actively’ ends a life-sustaining treatment.8 9

The terminology of ‘aktive Sterbehilfe’ and ‘passive Sterbehilfe’ has been criticised by numerous scholars because it conflates descriptive and normative distinctions relevant to end-of-life practices. In its statement ‘Self-determination and care at the end of life’, the German National Ethics Council (which was replaced by the German Ethics Council in 2007) not only provides a detailed critique of the ‘Sterbehilfe’ terminology but also proposes an alternative terminology to distinguish the different decisions and actions at the end of life.i 10 However, according to the authors' experiences, the distinction of ‘aktive Sterbehilfe’ and ‘passive Sterbehilfe’ is still most influential to the perception and practice of healthcare professionals.

The professional guidance on physicians' end-of-life practices issued by the German Medical Association indicates that any medical measure may be withheld or withdrawn if this is in accordance with the patient's will.11 In legal terms, however, there have been divergences between different verdicts by the courts with respect to the scope of limitation of treatment. It is only very recently that the German Federal Court of Justice (Bundesgerichtshof) has clarified that the withdrawal of life-sustaining treatment, such as artificial nutrition, is legal if this decision is based on the patient's will. Neither the current state of health of the patient nor the fact that the decision is based on the patient's present will or an expression of will made by a patient before losing legal capacity poses limits to such end-of-life practice.12

In contrast to many other countries,13–15 there is a scarcity of empirical data on the end-of-life practices performed by physicians in Germany. The available data are either older16 17 or are difficult to interpret because of methodical weaknesses such as the use of the aforementioned ‘Sterbehilfe’ terminology as part of the survey instrument.18 In addition, the data which have been elicited in a number of European countries—for example, as part of the EURELD study (in which Germany did not participate)—cannot easily be transferred to the situation in Germany because of the well-known country-specific differences with respect to end-of-life practices.1 15 Detailed information about end-of-life practices can make a contribution on different levels. First of all, and relevant to the public discourse on an appropriate framework for normative and practical aspects of end-of-life decision-making in modern medicine, such data can provide insight regarding the current state of practice and the findings may direct the focus of interest to aspects which deserve further attention on a normative and/or practical level. A second contribution of empirical-ethical research on end-of-life practice is that it can provide a starting point for normative reflection and, at the same time, inform ethical analysis applied to the field of modern medicine. While there is currently a considerable debate about the appropriate relationship between norms and empiricism in the context of applied ethics, there seems to be a widely shared assumption that any applied normative analysis is in need of information regarding the empirical aspects of the issue at stake.19

In this paper we present and analyse findings of a survey on practices regarding the limitation of medical treatment at the end of life as these were reported by physician members of the German Society of Palliative Medicine by means of the German version of the EURELD questionnaire. While data on the overall incidence of end-of-life practices will be published elsewhere,20 this research focuses on a detailed description and interdisciplinary analysis of ethically and clinically relevant aspects of decisions about the limitation of treatment at the end of life.


A postal cross-sectional survey of end-of-life practices was conducted among all physician members of the German Society for Palliative Medicine (N=1645). In line with the interdisciplinary approach of this research project, representatives from normative as well as empirical disciplines including clinical practitioners were involved during all stages of the project. One of the main tasks was to ensure that ethical as well as socio-empirical factors were taken into account. In addition, clinicians formed part of the research group to facilitate eliciting and interpretation of data by consideration of the perspective of respondents. In line with the research proposal accepted by the research ethics committee of the Medical Faculty, Ruhr-University Bochum, the contact details of all the physician members of the German Society for Palliative Medicine were provided by the secretary of the society as address labels. For the protection of anonymity, there was no identifying code on the questionnaire. The whole sample received the questionnaire for the first time in February 2009 and a second time in March 2009 as a reminder of the study. In the cover letter of the follow-up reminder, potential participants were informed that they should respond only once.

The German version of the EURELD questionnaire, which had already been used in the German-speaking part of Switzerland, was used as a survey instrument.15 Box 1 summarises the key questions of the survey instrument relevant to this research.

Box 1

Key questions on physicians' practices of limitation of treatment at the end of life


Following the procedure described by Seale,14 potential participants were informed on the first page of the questionnaire that all the questions of the survey refer to the patient who had most recently died under their care. Participants in the study who indicated on the cover page of the questionnaire that they had not cared for a dying patient within the last 12 months or that they had worked abroad during the last year were asked to return the questionnaire without any further information.

Data were collected in a SPSS data file. Free text comments regarding the type of limited treatment was categorised according to earlier research by Bosshard et al1 21 as part of the EURELD study. Descriptive and statistical analysis to test for possible associations between practices of limitation of medical treatment characteristics from the point of view of the patient, the physician and contextual factors was conducted based on the findings of earlier empirical research21–23 and analysed using SPSS Version 18.0 for Windows; p values <0.05 were considered significant.


Nine hundred and one questionnaires were returned following the initial mailing of questionnaires and one follow-up. In addition, 31 questionnaires were returned to sender because the physician could not be located at the address given. The response rate was 55.8%. One hundred and fourteen respondents (12.7%) indicated that they had either not treated a patient who had died within the last 12 months (N=84) or that they had not worked in Germany during this time period (N=30). These questionnaires were excluded as well as seven other questionnaires in which further analysis was not possible due to a significant lack of data. In total, 780 questionnaires were eligible for further analysis, 447 (57.3%) male respondents and 332 (42.6%) female respondents with one case where the gender was not indicated.

Participants reported limitation of treatment in 69.1% of all cases; in 60.6% of cases treatment was withheld with foreseeable or intended shortening of life and in 44.2% of patients the treatment was withdrawn. The decision to limit treatment most often involved artificial nutrition (19%), chemotherapy (14%), antibiotics (11%) and medication other than antibiotics (11%). The proportions in which treatment was withheld or withdrawn differed depending on the measure in question. While there were more cases of withdrawal than withholding of treatment in the case of medication and catecholamines, the proportion of cases in which treatment was withheld was larger than those in which treatment was withdrawn with respect to limitation of nutrition, chemotherapy antibiotics, dialysis, transfusions and respiration. Table 1 summarises the frequencies of different types of medical treatment which were limited and provides an overview of the proportions withheld or withdrawn in these cases.

Table 1

Frequency of types of treatment limitation

Due to the structure of the survey instrument, the estimated time of shortening of life as a consequence of limitation of medical treatment could only be determined in cases in which this decision was most relevant according to a predefined hierarchy in the questionnaire. In 75 of 234 eligible cases there was no estimated shortening of life, in 95 cases the life-shortening effect was estimated to be <7 days and in 17 cases the life-shortening effect as a consequence of withholding or withdrawal of treatment was estimated to be >1 month. Table 2 summarises the results with respect to the estimated shortening of life as a result of withholding and/or withdrawal of treatment.

Table 2

Estimated shortening of life as a consequence of limitation of treatment*

In 84 cases the respondents indicated only one limited intervention. In these cases it was possible to correlate the measure limited at the end of life with the estimated time span of shortening of life. Most frequently, physicians estimated that the limitations of treatment shortened the life of the patient by <7 days (33/84 cases). Moreover, in 29 cases the respondents indicated that there was no life-shortening effect at all due to the limitation of a defined treatment. In six cases it was reported that the limitation of treatment might have had a life-shortening effect of >1 month. As indicated in table 3, there were different patterns of expected shortening of life depending on the measure in question. Life-shortening as an effect of limitation of treatment was expected in all cases of transfusions and in all but one case of limitation of dialysis (5/6 cases) and respiration (6/7 cases). In contrast, in more than half of the cases in which nutrition was limited (7/12 cases), this action was considered to have no life-shortening effect. The limitation of antibiotics and other medications as well as the limitation of hydration was most commonly associated with expected life shortening of <7 days while, in cases of limitation of chemotherapy and respiration, it was estimated that there was a life-shortening effect of >1 week in more than one-third of cases.

Table 3

Estimated shortening of life as a consequence of limitation of treatment*

Based on the results of the descriptive analysis (table 4) and the findings of earlier studies,1 22 we explored possible determinants for limitation of treatment with reference to patient characteristics, physician characteristics and contextual factors.

Table 4

Types of limitation of treatment, descriptive analysis according to patients' and physicians' characteristics and contextual factors

Bivariate statistical analysis indicated that the withholding of treatment resulting in a foreseeable life-shortening effect was performed significantly more frequently for patients aged ≥65 years than for younger patients (64.6% vs 56.1%, p=0.019). Treatment was also withdrawn significantly more frequently from women than from men (52.5% vs 43.4%, p=0.014). There was no significant association between place of death (hospital (72.6%) vs others (72.3%), p=0.925 or home (69.2%) vs others (73.8%), p=0.202) and incidence of limitation of treatment, nor was there an association between physicians' qualifications (defined as an additional certificate in palliative medicine) and the incidence of limitation of treatment (physicians with a qualification in palliative medicine limited treatment in 74.0% of all cases vs 68.3%, p=0.169). Further bivariate analysis indicated that, depending on the disease indicated by the respondents as the underlying condition leading to death, there were different patterns of limitation of medical treatment. Physicians reported less withholding of medical treatment with possible life shortening in cancer patients than in patients with other diseases (57.6% vs 71.9%, p=0.001). In addition, physicians reported withdrawal of medical treatment with the intention to shorten life in 12.8% of all cancer cases compared with 20.2% of patients with other diseases (p=0.016). In patients with diseases affecting the nervous system, treatment was significantly more often withheld with the intention to shorten life than in patients with other diseases (41.5% vs 21.3%, p=0.001).


In this paper we present a detailed analysis of the practices of limitation of medical treatment reported by physician members of the German Society for Palliative Medicine. The analysis is part of a larger research project on end-of-life practices of physician members of the German Society for Palliative Medicine, the first results of which have been published elsewhere.20

The potential contribution of a survey to the interdisciplinary discourse on empirical and ethical aspects of end-of-life practices in our view depends very much on the collaboration between researchers in different fields during the different stages of the research process. The term ‘empirical-ethical study’ as used in this study refers to the fact that the aims of this survey and the methods used to gather the respective data are informed by normative concepts (eg, distinctions between different end-of-life-practices). In addition, selected findings are linked to ethical debates (eg, frequency of treatment limitation associated with age and age-based rationing). However, a thorough normative analysis of the related normative issues is beyond the scope of this paper. Following key principles of the concept of critical applied ethics as described by Leget et al, one crucial task of the interdisciplinary research team was to ensure that, during the design phase of study as well as data interpretation, the different disciplinary perspectives were brought together and reflected on critically.23 One concrete step in this respect was the selection of the survey instrument. An important reason for the decision in favour of the EURELD questionnaire is the terminology used to distinguish different end-of-life practices (see box 1). The survey instrument defines limitation of treatment either as ‘withholding’ or as ‘withdrawal’ of medical treatment. In addition, the survey instrument captures normative relevant categories such as the distinction between foreseeable versus intended shortening of life. A terminology which, on the one hand, reflects ethically relevant concepts and, on the other hand, captures the perspective and experiences of respondents whose practice is to be explored is an important criterion for sound empirical-ethical research.24 25 In addition to the methodical challenges of the survey instrument which have been discussed elsewhere,14 20 one limitation of this study is associated with the sample—namely, physician members of the German Society for Palliative Medicine—which is not representative of physicians working in Germany. The results therefore cannot easily be compared with data from other studies such as those published by the EURELD Consortium.1 15 Further methodical limitations comprise possible selection bias, socially desirable answers and problems with recall.

The findings of this study confirm that physicians are frequently confronted with decisions about treatment limitation and associated clinical and ethical challenges.1 2 This finding is relevant to the current ethicolegal debate insofar as it often focuses on the normative issues related to ending a patient's life on request or physician-assisted suicide. This study not only indicates that decisions on the limitation of treatment are part of the day-to-day practice, but also provides insight into the spectrum and the clinical and ethical complexities of treatment limitation. In the majority of patients decisions had to be made about the limitation of more than one treatment. Further analysis of the data suggests that there are different patterns depending on the type of limited treatment and the estimated effect on shortening of life (tables 2 and 3). There are also significant associations between the incidence of limitation of treatment leading to shortening of life and the disease being treated. While these results need to be interpreted cautiously in view of the fact that three-quarters of the patients in this study had cancer, other explanations for the disease-related differences may be varying perceptions and also the availability of treatment for different conditions at the end of life. Given that cancer is at the centre of clinical research interest worldwide and attracts specific attention in the public debate, there may be a shared understanding that further treatment options are available, even in the late phase of the disease. Another explanation for the less frequent use of treatment limitation in patients with cancer was identified as part of a qualitative study conducted by two of the authors (JS, JV). According to this research, oncologists viewed limitation of cancer treatment as a practice which does not shorten life because, in many cases, the continuation of treatment would put patients' lives at risk.25 These and other findings illustrate the many facets of clinical cases which are summarised under the label ‘limitation of treatment’ and which, at the same time, differ with respect to ethically relevant characteristics.

Information about the frequency and characteristics of different end-of-life practices as elicited in this study can contribute to the normative debate by directing the focus of interest to challenges which are part of the day-to-day practice at the end of life. Another not completely separate function of such research is to elicit data which can serve as a starting point for further ethical analysis. One interesting result in this respect is that there was significantly more withholding of medical treatment in patients aged ≥65 years than in younger patients. This finding has been published previously21 and has been used as a starting point for debates on whether so-called ‘age-based rationing’ could be justified on normative grounds. While an ethical analysis of the argument around age-based rationing is beyond the scope of this paper, we would like to take the empirical finding as an example to illustrate the need to analyse critically the content of data generated by a survey and the relevance of such empirical analysis for an applied ethical analysis. Looking at qualitative research regarding the role of age in treatment decision-making, there is evidence that age is often used to refer to the fitness of the patient or the social relationships and obligations of a patient.23 26 27 There is quantitative as well as qualitative research which indicates that, independent of other factors, younger patients are treated more aggressively during the last phase of life than older patients.21 26 This more differentiated analysis of the possible role of age may serve as an example of the relevance of a solid empirical analysis prior to the use of such data in normative debates. It may well be that the applied ethicist following such analysis decides to focus on the normative issues related to ‘maximal therapy’ of younger patients rather than on the question of age-based rationing.

Similar to the findings of earlier research,21 physicians in this study performed more withholding of treatment than withdrawals (60.6% vs 44.2%). Part of this result may be explained by the nature of some of the interventions (eg, hospital admission, surgery). However, as discussed in the introduction to the paper, the finding on physician-reported actions also fits well with results on physicians' perceptions and views, where the withdrawal of treatment is perceived to be psychologically more difficult and also to be illegal.7 28 A practice in which physicians would rather not choose to initiate a certain medical treatment than to withdraw it after it had been shown to be ineffective would be ethically problematic, not least because this practice would deny patients a chance of treatment in situations in which it is known that only a small percentage would profit from such treatment.29 This situation is clinically relevant in so far as there is often limited evidence on the outcome of, for example, new and innovative ‘last-line’ treatment, which means that a limited phase of trying such treatment followed by an evaluation and possible withdrawal seems to be the most appropriate approach.

Based on these findings, we suggest further empirical-ethical research on end-of-life practices for at least two reasons. First, we would argue that ethically informed empirical research, especially qualitative studies, could give a better understanding of the interaction of medical and value judgements at the end of life. Second, we believe that conducting empirical-ethical studies offers an excellent opportunity to provide a reflective account on the ‘doing of empirical ethics’. Such a twofold approach to further empirical research in medical ethics can contribute to an empirically informed normative analysis, as it may be able to enrich the current discussion on the conceptions of empirical ethics.


We would like to thank all participants for their time and consideration in taking part in this study. We thank Dr Georg Bosshard, Dr Susanne Fischer, Dr Agnes van der Heide and Professor Clive Seale for their support in preparing the study and the reviewer for the constructive comments on an earlier draft of this paper. We would also like to thank Stephanie Dahm for data checking and Dr Thomas Schindler for his support on behalf of the secretary of the GSPM.


  • JS and JH contributed equally to this paper.

  • Funding JH was funded by a start up grant of the Ruhr University Bochum (“Programm zur Anschubfinanzierung des wissenschaftlichen Nachwuchs”). JS and AB form part of the NWR-Junior Research Group “Medical Ethics at the End of Life: Norm and Empiricism” at the Institute for Medical Ethics and History of Medicine, Ruhr-University Bochum which is funded by the Ministry for Innovation, Science, Research and Technology of the German state of North Rhine-Westphalia.

  • Competing interests CM-B (past president) and JS are members of the German Society for Palliative Medicine.

  • Ethics approval This study was conducted with the approval of the Research Ethics Committee of the Medical Faculty of the Ruhr-University Bochum.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • i The suggested terms are ‘end-of-life care’, ‘therapy at the end of life’, ‘letting die’, ‘assisted suicide’ and ‘killing on request’.


Relevant Article

Free sample
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of JME.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.