Article Text

Download PDFPDF

The role of law in decisions to withhold and withdraw life-sustaining treatment from adults who lack capacity: a cross-sectional study
  1. Benjamin P White1,
  2. Lindy Willmott2,
  3. Gail Williams3,
  4. Colleen Cartwright4,
  5. Malcolm Parker5
  1. 1Australian Centre for Health Law Research, Queensland University of Technology, Brisbane, Queensland, Australia
  2. 2Australian Centre for Health Law Research, Queensland University of Technology, Brisbane, Queensland, Australia
  3. 3School of Population Health, University of Queensland, Brisbane, Queensland, Australia
  4. 4ASLaRC, Southern Cross University, Tweed Heads, New South Wales, Australia
  5. 5University of Queensland, Brisbane, Queensland, Australia
  1. Correspondence to Professor Benjamin P White, Australian Centre for Health Law Research, Queensland University of Technology, GPO Box 2434, Brisbane, QLD 4000, Australia; bp.white{at}


Objectives To determine the role played by law in medical specialists' decision-making about withholding and withdrawing life-sustaining treatment from adults who lack capacity, and the extent to which legal knowledge affects whether law is followed.

Design Cross-sectional postal survey of medical specialists.

Setting The two largest Australian states by population.

Participants 649 medical specialists from seven specialties most likely to be involved in end-of-life decision-making in the acute setting.

Main outcome measures Compliance with law and the impact of legal knowledge on compliance.

Results 649 medical specialists (of 2104 potential participants) completed the survey (response rate 31%). Responses to a hypothetical scenario found a potential low rate of legal compliance, 32% (95% CI 28% to 36%). Knowledge of the law and legal compliance were associated: within compliers, 86% (95% CI 83% to 91%) had specific knowledge of the relevant aspect of the law, compared with 60% (95% CI 55% to 65%) within non-compliers. However, the reasons medical specialists gave for making decisions did not vary according to legal knowledge.

Conclusions Medical specialists prioritise patient-related clinical factors over law when confronted with a scenario where legal compliance is inconsistent with what they believe is clinically indicated. Although legally knowledgeable specialists were more likely to comply with the law, compliance in the scenario was not motivated by an intention to follow law. Ethical considerations (which are different from, but often align with, law) are suggested as a more important influence in clinical decision-making. More education and training of doctors is needed to demonstrate the role, relevance and utility of law in end-of-life care.

  • Care of the Dying Patient
  • End-of-life
  • Ethics
  • Law
  • Living Wills/Advance Directives

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Law is part of the regulatory framework that governs end-of-life care; most developed nations have laws that facilitate decisions about withholding and withdrawing life-sustaining treatment (WWLST) from adults who lack capacity.1–3 For example, the frameworks generally allow adults to complete advance directives (ADs) and appoint substitute decision-makers for potential loss of capacity. These laws aim to safeguard patient interests (including autonomy interests), provide certainty and protection for doctors acting within the law and establish a process for resolving otherwise intractable disputes.4 Failure to know the law and follow it can place doctors at legal risk and also risks patients' rights and interests.5 ,6 But what role does the law actually play in decisions about WWLST for adults who lack capacity? Do doctors know the law and does it influence their decision-making?

There is a small but growing body of research that demonstrates legal knowledge gaps of doctors (or students) in this area in the UK,7–9 Australia5 ,10–12 and the USA.6 ,13–18 A few studies have also determined that perceptions of legal risk can alter medical decision-making at the end of life.13 ,19 ,20 One set of studies by McCrary and others has linked legal knowledge with attitudes, namely whether doctors are likely to be legally defensive.13 ,14 ,21 But little is known about the role of law in medical decision-making, so far as what it actually states and requires. And here we distinguish this from what doctors might perceive the law requires of them (rightly or wrongly) and also from the impact that more generalised concerns about legal risk may have. Research based on actual legal knowledge requires that knowledge to be measured and then assessed in relation to how that affects practice, including compliance with the law.

There is also a substantial body of research that points to challenges with doctors following legal mechanisms for decision-making such as ADs.20 ,22–26 But again, this research does not grapple with what the law actually requires as a normative force on decision-making. For example, it is unclear whether doctors follow ADs because they are required to do so by law or because of the weight they give to patients’ wishes (or because of both reasons).

This paper addresses these gaps by exploring medical specialists' knowledge of, and compliance with, relevant legal frameworks and the role of law in their decision-making about end-of-life care. The central research question of this study is: Are doctors who know the law more likely to decide to follow it?


A postal survey of medical specialists who practised in acute care from the three largest Australian states of New South Wales, Victoria and Queensland was conducted. Queensland has substantially different law from the other two states; it is inconsistent with international trends (ADs can be overridden on the basis of good medical practice)27 and so has been excluded from the results presented in this paper which focus on AD compliance. The survey instrument, developed over 18 months, was informed by a detailed review of the law in each state,27–29 focus groups, pretesting and piloting with doctors. The accuracy of the legal questions and responses were confirmed by independent legal experts. A more detailed description of the development of the survey instrument, and the wider project methodology, has been published elsewhere.30

The sample comprised all specialists who: (1) identified their main specialty as being in emergency, geriatric, palliative, renal and respiratory medicine, medical oncology or intensive care and (2) were on the AMPCo Direct database31 at the time the instrument was distributed (total N=2858 for three states; New South Wales and Victoria=2222). AMPCo Direct has Australia's most comprehensive and accurate doctor database and has been used in other major studies of Australian doctors.32 These specialties were chosen as they are likely to be involved in decisions about WWLST. This was determined by a review of relevant literature, interviews and an analysis of pilot results.

AMPCo Direct administered the survey mail-out from July 2012. Strategies to improve response included having the survey instrument professionally designed, providing incentives (continuing professional development points, educational material and a chance to win one of six bottles of prestige wine), engaging with the colleges and societies of target specialties and publishing editorials in relevant professional journals to request participation in the study.33 ,34 Two follow-up requests were sent to non-responders and the survey closed on 31 January 2013.

This study was approved by the Human Research Ethics Committees at the Queensland University of Technology (1100001137), the University of Queensland (2011001102) and Southern Cross University (ECN-11-222). All participants provided informed consent before taking part in the study. This paper was drafted using the Strengthening the Reporting of Observational Studies in Epidemiology Statement (see online supplementary file 2).


The survey instrument had six sections: perspectives about the law; education and training; knowledge of the law; practice of and compliance with the law; experience in making end-of-life decisions and participant characteristics.

Perspectives about the law were examined through two questions where respondents were asked to rate, on a five-point scale from strongly disagree to strongly agree, with unsure as the middle option, their level of agreement with a series of statements (see online supplementary file 1: figure S1). One question contained 11 statements about the role of law in this area of medicine, while the other question contained 1 statement about specific aspects of knowing and following the law. Responses to each of these questions were analysed separately to generate a score representing how positive each respondent's attitude was to the role of law, and to knowing and following the law, which were then examined in light of specialists' compliance with law.

Compliance with law was measured using a scenario involving a patient who had completed an AD (jurisdiction-specific terminology used) 5 years previously, soon after being diagnosed with AIDS. In his AD, he refused antibiotics for any future life-threatening infection and wished only to be kept comfortable. He becomes ill with a life-threatening infection and now requires antibiotics to survive. Both his family and doctors wish him to receive antibiotics as he would be likely to recover from the infection and continue to live as before. If he is not given the antibiotics it is likely he will die. Respondents were asked whether they would treat or not (the latter being in compliance with law in both states) and they were then asked to select from a predefined list (with ‘other—please specify’ available) the option that best described the reason for their decision. Respondents were also asked to rate the relevance to their decision in this scenario of 11 items on a four-point scale from not relevant to very relevant.

The knowledge section contained two questions with a total of seven items. The first question comprised six items with answers being true, false or do not know in relation to relevant state law (do not know counted as incorrect). Four items specifically assessed knowledge of ADs and one of them (the ‘key knowledge question’ or KKQ) was particularly relevant to the compliance scenario: ‘The law requires you to follow a valid AD that refuses life-sustaining treatment even if providing that treatment is clinically indicated’. The second question in the knowledge section involved a scenario asking which of four plausible decision-makers had legal authority to make medical decisions for an adult patient without capacity?

Respondents were then divided into four groups according to their response to the AD scenario and their response to the KKQ. This resulted in ‘knowledgeable’ or ‘accidental’ compliers—compliers who correctly answered the KKQ or not, respectively—and ‘knowledgeable’ or ‘accidental’ non-compliers—non-compliers who correctly answered the KKQ or not, respectively.

Statistical analysis

Questionnaires were coded and double-entered into an Access database and transferred to SPSS V.20 (IBM) and SAS V.9.3 (SAS Institute) for analyses. These analyses were conducted on the combined New South Wales and Victorian data. Preliminary analyses examined descriptive statistics and bivariate associations between categorical variables by χ2 tests. Predictors of compliance (not treating/treating) were assessed using multivariable logistic regression, with likelihood ratio tests used for assessing overall significance of covariates. Variables examined as predictors of compliance and knowledge were: state, age, gender, main specialty group, religion, years of practice, country of birth, country of degree and number of decisions made in relation to WWLST. Bonferroni adjustments were used to identify significance of differences among the seven specialty groups. Knowledge scores were analysed as means and SDs since, while only eight distinct scores could be attained, the overall distribution was approximately normal (see online supplementary file 1: figure S2). A two-sample t-test was used to compare attitude scores and ratings of reasons for the decision on the compliance scenario and knowledge and between compliers and non-compliers.

Principal components analyses were used to develop a score (one factor) related to respondents' attitudes to the role of law and importance of following and knowing the law. Logistic regression was then used to examine the relationship between compliance with the law and attitudes to the role of law and importance of following and knowing the law.

Principal component analysis followed by Varimax rotation was applied to the 11 items rated for their relevance to the decision on the compliance scenario, to obtain four factors, using the Kaiser criterion for the number of factors. Items with loadings of at least 0.4 were considered to be ‘important’ and used to interpret the rotated factors.35 The four compliance groups (described above) were compared on the four factor scores obtained using regression analysis, assuming a normal distribution of factor scores, with Nelson-Hu adjustments for multiple comparisons.

A two-sided α level of 0.05 was used to define statistical significance.


After excluding ‘return to sender’ questionnaires and a small number of paediatricians and trainees, we identified a denominator population of 2702 across the three states or 95% of the original mail-out sample (New South Wales 1147, Victoria 957, Queensland 598). The total response rate for New South Wales and Victoria was 31% (N=649/2104) with 29% (N=335/1147) from New South Wales and 33% (N=314/957) from Victoria. Response rate by specialty across these states ranged from 75% (n=21) for palliative care specialists in Victoria to 22% (n=30) for medical oncologists in New South Wales. A comparison of respondents by age, gender, specialty and state with the original AMPCo sample found that respondents were similar on most comparison variables except that there were fewer younger doctors among respondents than in the sample population (see online supplementary file 1: table S1).

Compliance with the law

There was a low level of compliance with law in the AD scenario, with only 32% (95% CI 28% to 36%) saying that they would follow the AD. There was a significant difference between the states: 28.3% and 36.9% of doctors in New South Wales and Victoria, respectively, complied with the law, with little variation over other characteristics (see online supplementary file 1: table S2). The strongest variable predicting legal compliance was specialty (likelihood ratio test (LRT), p=0.003, after adjusting for state, age, gender, country of birth, years of practice and religion) with specialists in palliative and geriatric medicine more likely, and respiratory medicine specialists less likely, than respondents from the other specialties to comply (table 1); adjustments made little difference to the variation among specialty. Comparisons which were significant after Bonferroni adjustment were palliative medicine specialists versus respiratory medicine specialists (p=0.05), and geriatric medicine specialists versus intensive care and respiratory medicine specialists (p=0.05 and p=0.03 respectively).

Table 1

Decisions that comply with the law

Role of knowledge and attitudes about law

Knowledge of law

Associations between knowledge of the law and compliance with it were also tested by looking at compliance in light of responses to the seven legal knowledge items in the survey and the four items that specifically related to ADs. There was a significant relationship between compliance and legal knowledge in that those medical specialists who scored better on the seven general knowledge items and the four AD knowledge items were more likely to follow the law. Of note is that almost all of this significance was due to one item which is directly relevant to the scenario (the KKQ): ‘The law requires you to follow a valid AD that refuses life-sustaining treatment even if providing that treatment is clinically indicated’ (table 2).

Table 2

Association between knowledge of and compliance with law

Attitudes to the role of law in medicine and knowing and following it

Having a positive attitude to the role of law did not predict following law in the AD scenario. Also, no significant relationships emerged between respondents' attitudes to statements about knowing and following the law at an abstract level and actually following the law in the scenario (see online supplementary file 1: table S3).

Reasons for decision

Table 3 shows the reasons respondents selected for why they would or would not follow the AD (and therefore law) in the scenario. This table is divided into whether respondents correctly answered the KKQ or not, which would be expected to predict compliance. This allows a comparison of the reasons that knowledgeable respondents chose for their decision with the reasons chosen by those who were less knowledgeable. There was no association between the reasons for following the AD (and law) and legal knowledge (χ2 test, p=0.83), but there was a marginally significant association between the reasons for not following the AD (and so not following the law) and legal knowledge (χ2 test, p=0.050). Medical specialists who did not follow the law and chose the reason ‘I do not have to follow the AD because it is inconsistent with what is clinically indicated’ (n=28) were less likely to be legally knowledgeable. This result is not unexpected as that reason directly contradicts the correct answer to the KKQ, which is that ADs prevail over what is clinically indicated.

Table 3

Why respondents comply with law or not by knowledge of KKQ: n (%)

Relevant variables in decision-making

To help understand reasoning for decisions in the AD scenario, specialists were asked to rank the relevance of 11 items in their decision-making (table 4). ‘Following the law’ was the fifth most important item and its mean fell between relevant and somewhat relevant (on a four-point scale also comprised very relevant and not relevant). When examined separately for those who complied with the law and those who did not, ‘following the patient's AD’, ‘following the law’, ‘hospital policies’ and ‘concerns about being sued or criminal prosecution’ were more relevant for those who complied with the law, and ‘patient's expected quality of life after proposed treatment’, ‘whether treatment is clinically indicated’, ‘your personal ethical principles’, ‘family views’ and ‘views of your colleagues about what you should do’ were more relevant for those who did not comply with the law (table 4).

Table 4

Mean scores for relevance of items to decision-making, by whether respondents comply with law or not (significance bolded)

A factor analysis of these 11 items yielded four independent factors: patient-specific considerations (32% of common variance explained); professional guidance (26%); fear or respect for law (25%) and personal beliefs (17%), which together explained 65% of the total variance in the 11 items. All items had important loadings on at least one of the four factors (table 5, see online supplementary file 1: table S4). ‘Following the patient's AD’ loaded negatively on ‘patient-specific considerations’ (which was negatively correlated with each of the other three variables with important loadings on this factor) and positively on ‘fear or respect for law’.

Table 5

Factors (loadings) obtained from items of relevance in decision-making for items with important loadings, and per cent of explained common variance for each factor

Four factor scores were then calculated for each respondent, as weighted means (weights=loadings in table 5) of the 11 items (see online supplementary file 1: figure S3). These represent the extent to which the respondent indicated each of the factors was relevant to his or her decision-making. The mean of each of these ‘relevance scores’ was calculated for the knowledge/compliance subgroups defined by knowledge of the law assessed by the KKQ and compliance with the law (figure 1).

Figure 1

Relevance of items in decision-making by knowledge and compliance, with 95% CIs.

Variation among knowledge/compliance subgroups in relevance scores for ‘patient-specific considerations’ and ‘fear or respect for law’ was statistically significant (p<0.001 in both cases), while variation among the subgroups in ‘professional guidance’ and ‘personal beliefs’ was not.

For ‘patient-specific considerations’, all mean scores were significantly different from the overall mean score of zero. For ‘fear or respect for law’, all mean scores except those for knowledgeable non-compliers were significantly different from zero. For both factors, and within compliers and non-compliers, the difference between those with and without knowledge of the law was not significant.

This means that fear or respect for the law was most relevant for those who complied with the law and patient-specific considerations were least relevant, the latter heavily reflecting clinical considerations. For those who did not comply with the law, the opposite occurred (figure 1). However, these two factors were independent of knowledge as measured by the KKQ. Professional guidance and personal beliefs were not associated with either compliance or knowledge.


Principal findings

There was a low rate of compliance with law in the AD scenario. Where following the law was inconsistent with what is clinically indicated, specialists' prioritised patient-related clinical factors. Although there was an association between knowledge of the law and legal compliance, further analysis of the reasons for decision-making and the matters specialists considered relevant revealed that knowledge of the law did not affect decision-making in the scenario. Legally knowledgeable specialists and those who were not knowledgeable complied with law (by not treating the patient) for the same reasons and relying on the same factors. Likewise, level of legal knowledge did not affect how specialists who did not comply with the law (by treating the patient) made their decisions. This is particularly interesting in the case of those who are giving more weight to the law but do not know what it requires. Accordingly, what the law requires (and here we distinguish this from doctors' generalised concerns about law or perceptions about what the law might be) is not an influential factor in decision-making about life-sustaining treatment for adults who lack capacity.

Explanations and implications

We suggest that there is some other variable operating which is associated with enhanced knowledge of the law, and that it is responsible for the association between specialists' compliance and legal knowledge. A logical candidate is ethical considerations. The interconnectedness of law and ethics means that although they are distinct institutions, they help shape each other,36 and both law and autonomy-focused ethics point to not treating in the AD scenario. Medical ethics and law are also often taught together in an integrated way37 and so more legally informed specialists are likely to have had more instruction in ethics. Therefore, we suggest that it may be ethical considerations that are influential, and not law, for specialists who, in accordance with the AD, are not treating, but that these specialists also happen to be legally knowledgeable. This could also explain how a doctor acting on the basis of autonomy-focused ethical considerations could comply with the law (given they often overlap) without knowing the law.

There is some support for this hypothesis in the data. When asked to select reasons for not providing treatment in the AD scenario, following the patient's wishes featured significantly more highly than following the law (respondents could also choose that both considerations were equally important). Further, when looking at the variables specialists identified as relevant or not to their decision-making in the scenario, ‘your personal ethical principles’ was the third most selected option behind ‘patient's expected quality of life’ and ‘whether treatment is clinically indicated’. After that came ‘following the patient's AD’, and only then came ‘following the law’. This hierarchy of ethics then law is also supported by an exploratory study of emergency medicine specialists' use of ADs in decision-making.24

Further research is needed to better understand the respective roles of law and ethics (and their intersections) in specialists' decision-making in end-of-life care. Both of these social institutions purport to regulate these decisions but do so in different ways, for example, drawing on different sources and with different consequences for non-adherence.38 On one view, law is a binding set of norms imposed externally which brings with it the prospect of sanction for non-compliance. Law is also perceived by some to be a relatively stable set of standards that applies to all in the same way. By contrast, although there are widely accepted standards of conduct imposed by ethical principles such as those contained in professional codes, the duties imposed by ethical reasoning and how those duties are discharged in end-of-life care can vary depending on an individual doctor's views. This has implications for how compliance with ethical principles can be determined as ethical reasoning may legitimately point to more than one appropriate course of action.

Other research and limitations

Although ADs can influence treatment that patients receive at the end of life,39 our findings accord with other studies that show low rates of compliance with ADs that are inconsistent with what is clinically indicated.20 ,22 ,24 ,25 However, this study is significant because it explores whether accurate legal knowledge improves compliance, and the role that the law plays in specialists' decision-making. This enables us to understand whether non-compliance is due to a lack of knowledge or whether it is because other competing considerations are more important. Our results suggest that it is, at least in part, the latter. Despite law purporting to be a binding force, it does not appear to have significant influence on medical deliberations in end-of-life care. These findings have implications for medical education and indeed the role of law in medical practice generally.

A limitation of this research is low response rate (31%). This is a feature common to survey research involving doctors as response rates from this cohort are low and declining.40 ,41 While non-response bias cannot be ruled out, comparisons of respondents with the wider sample support the broad representativeness of those who participated in this research (see online supplementary file 1: table S1). Any potential bias in this study may have the effect of overestimating the relevance of law in this area of medicine. Non-responders are less likely to be legally knowledgeable and perhaps also less likely to be influenced by law, so the role of law may in fact be less than the limited one we found.

The sample, which included all doctors from the seven specialties most likely to be involved in end-of-life decision-making in the largest Australian states, is more representative than previous related studies which have generally been drawn from those participating in specified training courses or cohorts,7 ,9 ,20 specific health facilities13 ,14 ,21 ,25 or a single specialty or society.11 ,12 ,18 ,19 ,22 ,24 However, our results from seven specialties in the acute setting may not be generalisable to other doctors or other settings. The role played by law may also vary in other jurisdictions without comparable legal systems.

Another limitation is that our measurement of compliance is based on a particular scenario. Different results could occur where following the law is not clinically challenging, but a scenario where law and medicine are in conflict was used so that the impact of law on clinical decision-making could be better evaluated. A single scenario is also not able to test the full range of legal issues that can arise at the end of life.20


The limited role played by law in decisions to WWLST from adults who lack capacity is troubling. Not complying with law can result in patients being denied legal rights and place doctors at legal risk.5 ,6 Law, which is ultimately a reflection of community values, has an important role to play in medicine. A societal decision through the institution of Parliament (and sometimes the courts) has been made to establish decision-making processes that safeguard the rights and interests of a vulnerable group in our community, adults who lack capacity and to allow people to express treatment preferences in advance and appoint substitute decision-makers. While we are conscious of the limits of law in this area,27–29 departures from law have significant implications for patients, clinicians and the community.

This research demonstrates that to improve compliance with law, increasing legal knowledge is necessary,5 but not sufficient. This points to the need for education that addresses what the law is and the law's rationale and arguments for complying with it. Our conclusions about the possible role of ethical considerations in legal compliance also suggest a need to be clear when teaching law and ethics that they are separate, although interacting, social institutions. Both law and ethics impose obligations to be followed, but what the law requires can differ from what is deemed as ethically appropriate, either broadly or within particular groups in society. There may also be merit in conceptualising legal knowledge and compliance as an ethical and professional responsibility given ethics has a more entrenched role and greater legitimacy within the profession. More education and training is needed to demonstrate the role, relevance and utility of law in end-of-life care.



  • Contributors BPW, LW, CC, MP and GW were responsible for the study concept and design, acquired the data and interpreted it. BPW and LW drafted the manuscript. GW, CC and MP critically revised the manuscript for important intellectual content. GW and CC were responsible for the statistical analysis. BPW and LW supervised the study. BPW, LW, GW, CC and MP are the guarantors. We also acknowledge the research assistance of Mark Eade.

  • Funding This study was funded by the Australian Research Council Linkage Projects scheme (project no. LP0990329) and the seven guardianship bodies who were partner organisations: New South Wales Civil and Administrative Tribunal, New South Wales Public Guardian, Office of the Public Advocate (Victoria), Victorian Civil and Administrative Tribunal, Queensland Civil and Administrative Tribunal, Office of the Public Guardian (Queensland) and Office of the Public Advocate (Queensland). The Australian Research Council had no further role in the study. The partner organisations provided in-kind support (as required by the relevant funding scheme) that included assisting in the study design and drafting of the survey instrument. They did not have access to the data nor were they involved in the interpretation of the data. They did, however, have an opportunity comment on an earlier version of this manuscript. We are particularly grateful to Dr John Chesterman from the Office of the Public Advocate (Victoria) for his comments. All authors are independent from the funders.

  • Competing interests None declared.

  • Ethics approval This study was approved by the Human Research Ethics Committees at the Queensland University of Technology (1100001137), the University of Queensland (2011001102) and Southern Cross University (ECN-11-222).

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