• Clinical ethics
• ethics committees/consultation
• newborns and minors
• paediatrics
• research ethics

Shared decision making has been officially supported by every major critical care organisation in the USA and Europe, the American Medical Association, the American Academy of Pediatrics, and others.1–3 Indeed, the first recommendation in the American College of Critical Care Medicine's guidelines for family support is that healthcare providers should partner with families in making difficult choices.4 In the USA and elsewhere, too often have doctors shied away from active participation in decision making due to a mistaken belief that value-laden decisions should be made only by patients or their surrogate decision makers.1

In order to provide optimal care, doctors must be willing and able to bear some, or at times the majority, of the burden when difficult decisions must be made. Indeed, shared decision making is best understood as a continuum ranging from patient-driven decision making, through an equal partnership with care providers, to fully physician-driven decision making.5 How the team of patient, family and providers balance the burdens and responsibilities of decision making should generally be driven by the wishes of the patient or the surrogate decision maker.

Empirical research supports the need for flexibility in decision making. Data from US studies show that most people want their doctor to be an active participant in value-laden decision making.6 ,7 Indeed, these studies report that about 1 in 10 people prefer that the doctor makes the major decisions, including the decision to continue life-prolonging therapies, or redirect care to palliative treatment without life-prolonging interventions (and about 1 in 20 want their doctor to make such decisions without even asking for their opinion first). Further, these studies show that only about 10% of surrogate decision makers prefer that they themselves make life-and-death choices without first getting a recommendation from the doctor. These data teach us that the overwhelming majority of patients and surrogate decision makers prefer to share decision making with their doctor, and that there are nearly an equal number of people at the extremes who either want to make decisions alone or want the doctor to make decisions for them.

Because healthcare providers should tailor their recommendations and decisions to the values expressed by the patient (or on behalf of the patient by others), blanket default options are generally not appropriate unless there is a clear ‘best’ option. For example, if a child is admitted to the hospital with an asthma exacerbation, the default is that the child will be given albuterol and steroids because these agents have been shown to improve symptoms. Alternatively, in the case of an infant born with hypoplastic left heart syndrome, studies show that parents as well as doctors are split between choosing surgery versus choosing palliative treatment without life-prolonging interventions.8 The default treatment in the former is appropriate because this represents the current standard of care; however a default option in the latter would be inappropriate because there is no clear ‘best choice’ in the care of these infants.

Similar to the case of infants born with hypoplastic left heart syndrome, there is no clear ‘best choice’ in the care of infants born near the border of viability. Some family/provider teams will opt for life-prolonging care, while others will choose to focus on palliative treatment without attempting to prolong the infant's life. As such, blanket defaults in such cases may be suboptimal. Haward, Murphy and Lorenz performed a random sample email-based surveyed in the US to assess whether presenting a default option in such cases would influence the choices participants stated they would make.9 Not surprisingly, they found that participants were significantly more likely to choose the default option, which is consistent with previous research as noted by these authors.

The most notable finding in the Haward study, however, is that there was no association between participants' stated preference regarding the level of involvement of the physician (as measured by Haward's paternalism scale), and whether the participant chose the default option. Put another way, how participants actually made choices was not associated with how they reported that they prefer to make choices.

Shared decision making relies on healthcare providers accurately judging the extent to which patients want doctors to share in the decision-making process,5 yet, as Curtis and Tonelli noted, simply asking patients or decision makers how they prefer to make difficult decisions may not provide sufficient information.10 The Haward study is perhaps the strongest support of this assertion. Determining how best to assist patients and families in making difficult decisions, therefore, remains in the realm of personal experience and preference since we lack data specifying how to make accurate assessments. If individual-stated preference for a preferred method of decision making is not associated with actual decision-making preference, then providers have an extremely difficult task judging how best to assist patients and families in making tough choices.

Haward et al argue that presenting default options compromises autonomy, however, this assertion does not follow from the data. What is clear is that merely giving people a piece of paper (or an email) stating the default option and asking what they choose could never qualify as informed permission. In contrast, if a physician (or healthcare team) sits with parents and discuss all options; explains that parents should feel free to make whatever choice they think is best, or employ any method of decision making they feel most comfortable with; is present (physically, intellectually, emotionally, and when appropriate spiritually) to provide support; and makes recommendations, or bears the major burden of decision making when parents prefer, then using default options would not necessarily be in conflict with optimal shared decision making. Indeed, one could as easily interpret Haward's findings as follows: people generally trust physicians and prefer that physicians take an active role in guiding them through the decision-making process.

Of note, Haward et al mistakenly equate default options with informed non-dissent. In their survey, the default option was based solely on random chance. In contrast, the choice made when using informed non-dissent is derived using the patient (or parents') values,11 and this method is employed only when doing so is consistent with the patient (or parents') preferences.5 When informed non-dissent is employed, the doctor must first learn the patient's values in order to judge what choice that individual would likely agree with. As such, informed non-dissent is in no way analogous to default options devoid of consideration for individual values. Similarly, when discussing framing, Feutner and colleagues limited their discussion solely to children whose condition is terminal and without hope for recovery.12 As such, blanket default options in cases where real choices are to be made (like whether to resuscitate a micro-premie) should not be confused either with informed non-dissent or appropriate framing.

Most people want recommendations from their healthcare providers. In some cases, like the child with asthma (above), default options will be appropriate. In other cases, when fully informed individuals make divergent choices, defaults likely confuse communication. Given that the overwhelming majority of patients and families reject a model of decision making in which providers refuse to bear shared responsibility for decisions, it would be misguided to argue that appropriate sharing compromises autonomy. Indeed, when the patient or surrogate decision maker prefers that the doctor bears the full burden of decision making, then doing so fully respects the patient's right to choose how decisions will be made. Clearly, any defaults that are not informed by the values of the individual patient require scrutiny. Haward's data raise significant concerns, however, because they suggest that judging where on the shared decision-making continuum patients feel most comfortable may be extremely difficult.