University of Otago Law Festschrifts
Last Updated: 31 May 2019
6 LIFTING OUR GAZE: AN EPIDEMIOLOGICAL APPROACH
TO MEDICAL REGULATION
INTRODUCTION Mark Twain once said:1
There is no such thing as a new idea ... We simply take a lot of old ideas and put them into a sort of mental kaleidoscope. We give them a turn and they make new and curious combinations.
The two Professors Skegg, Peter and David, gave generations of students rich facets of ideas to add to their mental kaleidoscopes. Peter Skegg as a health lawyer, an expert on the interface between law, ethics, and medicine.2 His brother, David Skegg, as an epidemiologist, focused on understanding patterns at the population level to improve the health of communities.
As a student of both professors, it is perhaps unsurprising that my career developed at the confluence of these two disciplines. This chapter offers a glimpse of some of the new patterns that emerge when the disciplines of epidemiology and medical law are entwined in “new and curious combinations”.3 The research findings below are built on knowledge acquired at the University of Otago, and subsequent collaborations with valued colleagues from the Universities of Auckland, Otago, Melbourne, Harvard, and Stanford.4 They focus largely on Australia and New Zealand, where I have undertaken much of my research work. However, they have significant international applicability.
* This chapter is adapted from my Doctor of Medicine thesis: M Bismark “Learning from
claims and complaints: an epidemiological approach to medical regulation” (2015) Otago University Research Archive <https://hdl.handle.net/10523/5879>.
My discussion is offered in four parts. In Part 1, I begin with an overview of the harm caused by those who are tasked with healing: the adverse events in healthcare that are now a leading cause of death in developed countries. I then turn to the medico-legal agencies that have developed in an effort to provide accountability for such harm—by way of compensation, explanation and sanction in the aftermath of harm—or to support prevention of further harm. Part 2 explores some common concerns about these medico-legal processes, including a perception that we may have achieved “the worst of both worlds”:5 a system of regulation that goes too far on one hand and on the other hand, does too little. Part 3 encourages us to lift our gaze beyond individual cases and anecdotes to take an epidemiologist’s view of medico-legal processes. And finally in Part 4, I present a summary of our research findings arising from this approach, and offer some concluding comments and recommendations.
1. MEDICO-LEGAL RESPONSES TO IATROGENIC HARM
1.1 First, Do No Harm
First, do no harm is a central tenet of medical practice.6 Since the time of Hippocrates, doctors have sworn: “I will use my power to help the sick to the best of my ability and judgment; I will abstain from harming or wronging any man by it.”7 This obligation of non-maleficence is central to medical ethics,8 and we even have a word—“iatrogenic”—to describe those illnesses that were brought forth by the healer.9
As the tremendous power of modern medicine to do good has grown,10 so too has its potential to cause inadvertent harm.11 Highly specialised surgeries, complex medical procedures, and interdisciplinary care require humility, teamwork,
5 T Stoppard Rosencrantz and Guildenstern are dead (Faber, London,
1967) at 52.
6 R Gillon “‘Primum non nocere’ and the principle of non-maleficence” (1985) 291(6488)
British Medical Journal 130.
7 G Lloyd Hippocratic writings (Penguin, United Kingdom, 1983) vol 451 at 67.
8 A Jonsen, M Siegler and W Winslade Clinical ethics: A practical approach to ethical decisions
in clinical medicine (7th ed, McGraw Hill Medical, New York, 2006); T Beauchamp and J Childress Principles of biomedical ethics (7th ed, Oxford University Press, Oxford, 2013).
10 J Le Fanu The rise and fall of modern medicine (Hachette, London, 2011).
11 A Gawande Complications: A surgeon’s notes on an imperfect science (Profile Books, London, 2010); A Merry and A Smith Errors, medicine and the law (Cambridge University Press, Cambridge, 2001).
and consistency in order to be delivered safely. Yet medicine has been slow to relinquish traditional values of autonomy, independence, and self-sufficiency.12 Rates of burnout and impairment among practitioners are worryingly high,13 the work of skilled and dedicated practitioners is jeopardised by unsafe systems,14 and speaking up about risks to patient safety remains difficult for many.15
The true magnitude of preventable medical harm associated with modern medicine was not fully appreciated until the 1990s, when researchers at Harvard quantified the proportion of patients in Utah, New York and Colorado harmed and killed by medical errors.16 Their findings, as reported in the landmark report ‘To Err is Human’,17 suggest that more people die in a given year as a result of medical errors than from motor vehicle accidents or breast cancer. Shocked by the findings, and concerned about the extent to which they might hold true in other healthcare systems, researchers sought to replicate the Harvard methodology in other countries, including New Zealand,18 Australia,19 Canada,20 and the United Kingdom.21 The results were sobering. Across the developed world, around one in ten patients experiences a prolonged hospital stay, or disability after discharge, resulting from medical care itself, rather than the underlying disease. Among these adverse events, around half are judged to be preventable.
12 A Gawande “Cowboys and Pit Crews” (Harvard Medical School commencement address 2011, Harvard Medical School, Massachusetts, May 2011).
13 S Bruce, H Conaglen and J Conaglen “Burnout in physicians: A case for peer-support” (2005) 35(5) Internal Medicine Journal 272.
14 A Merry and A Smith Errors, medicine and the law (Cambridge University Press, Cambridge, 2001).
15 T Faunce and S Bolsin “Three Australian whistleblowing sagas: Lessons for internal and external regulation” (2004) 181 Medical Journal of Australia 44.
16 E Thomas and others “Incidence and types of adverse events and negligent care in Utah and Colorado” (2000) 38(3) Medical Care 261; T Brenna and others “Incidence of adverse events and negligence in hospitalized patients: Results of the Harvard Medical Practice Study I” (1991) 324(6) New England Journal of Medicine 370.
17 L Kohn, J Corrigan and M Donaldson (eds) To err is human: Building a safer health system (National Academics Press, Washington (DC), 2000).
18 P Davis and others “Adverse events in New Zealand public hospitals I: Occurrence and impact” (2002) 115(1167) New Zealand Medical Journal 9.
19 R Wilson and others “The quality in Australian health care study” (1995) 163(9) Medical Journal of Australia 458.
20 G Baker and others “The Canadian adverse events study: The incidence of adverse events among hospital patients in Canada” (2004) 170(11) Canadian Medical Association Journal 1678.
21 C Vincent, G Neale and M Woloshynowych “Adverse events in British hospitals: Preliminary retrospective record review” (2001) 322(7285) British Medical Journal 517.
From these findings grew a burgeoning patient safety movement, focused on addressing problems with quality and safety in healthcare.22 This movement has created a plethora of tools for reporting, analysis, and investigation of healthcare quality and practitioner performance. It has also spawned interventions that span the continuum from highly technical efforts to re-engineer devices,23 through to broad efforts to reform the “club culture” of medicine.24
1.2 Regulatory Responses
As concerns about patient safety grew, questions were also being asked about whether the medical profession could be trusted to regulate itself.25 A series of high-profile inquiries into apparent failures of the profession to do so effectively turned a spotlight on the role of external legal and regulatory mechanisms. In New Zealand, the 1987 Cartwright Inquiry into unethical conduct at National Women’s Hospital led to sweeping reforms, including the world’s first legislated code of patients’ rights,26 the establishment of an independent network of patient advocates, and the appointment of a health complaints commissioner.27 In Australia, high-profile inquiries into the harm caused by errant doctors Jayant Patel28 and Graeme Reeves29 questioned the profession’s willingness or ability to deal with poor performance among its own. Major reforms included the establishment of the Australian Health Practitioner Regulation Agency and the introduction of mandatory reporting of concerns about the health, performance, and conduct of practitioners by other practitioners.30 Events in the United Kingdom had an impact too: as governments in New Zealand and Australia watched the inquiries
22 L Leape, D Berwick and D Bates “What practices will most improve safety?: Evidence-based medicine meets patient safety” (2002) 288(4) JAMA 501.
23 D Gaba “Anaesthesiology as a model for patient safety in health care” (2000) 320(7237) British Medical Journal 785.
24 C Dyer “Bristol inquiry: Bristol inquiry condemns hospital’s ‘club culture’” (2001) 323(7306) British Medical Journal 181.
25 H Clayton and K Webb “The benefits of a ‘right-touch’ approach to health care regulation” (2014) 19(4) Journal of Health Services Research & Policy 198.
26 PDG Skegg “A fortunate experiment? New Zealand’s experience with a legislated code of patients’ rights” (2011) 19(2) Medical Law Review 235.
27 S Cartwright The report of the cervical cancer inquiry (Government Printing Office, Auckland, July 1988).
28 H Thomas Sick to death: A manipulative surgeon and a health system in crisis - a disaster waiting to happen (Allen & Unwin, Crows Nest, 2007).
29 D O’Connor Review of the appointment, management and termination of Dr Graeme Reeves as a visiting medical officer in the NSW public health system (New South Wales Health Department, Sydney, 2008).
30 M Parker “Embracing the new professionalism: Self-regulation, mandatory reporting and their discontents” (2011) 18(3) Journal of Law and Medicine 456.
into deaths caused by Harold Shipman,31 and at the Bristol Royal Infirmary,32 they asked themselves: Could it happen here? Could it happen now?
Self-regulation carries with it an obligation to ensure the competence and trustworthiness of members of the profession.33 But these scandals, and others, suggested that perhaps the medical profession had been enjoying the benefits of a highly regarded profession without fulfilling all of the obligations attached to self-regulation. Health ministers and officials, acting on behalf of the public, responded with new measures designed to protect patients’ rights, provide public accountability, and achieve a better balance between professional freedoms and responsibilities.
By 2012, both Australia and New Zealand had established medico-legal systems that rested on three core pillars: national health practitioner regulation boards with strong lay representation, independent health complaints commissioners, and a commitment to compensating patients who are seriously injured by medical care. A key difference between the two systems is that, in Australia, tort law remains the primary source of compensation (though a new national injury insurance scheme in Australia is expected to cover serious treatment injuries).34 In New Zealand, compensation is provided through a no-fault insurer, the Accident Compensation Corporation (ACC).35
2. AIMING HIGH, FALLING SHORT
2.1 Regulation has Proven to be a Blunt Instrument
The New Zealand and Australian medico-legal systems are often held up as being at the forefront of medical regulation internationally.36 And indeed, they do offer many advantages over a traditional tort-based system, with its rigid,
31 D Smith “Not by error, but by design - Harold Shipman and the regulatory crisis for health care” (2002) 17(4) Public Policy and Administration 55; J Smith The Shipman Inquiry: Fifth report; safeguarding patients: Lessons from the past - proposals for the future (The Stationery Office, London, December 2004).
32 I Kennedy Learning from Bristol: The report of the public inquiry into
children’s heart surgery at the Bristol Royal Infirmary 1984-1995
Stationery Office, London, January 2002).
33 R Paterson The good doctor: What patients want (Auckland University Press, Auckland, 2012).
34 D Weisbrot and K Breen “A no-fault compensation system for medical injury is long overdue” (2012) 197(5) The Medical Journal of Australia 296; Productivity Commission Disability care and support (Canberra, 31 July 2011).
35 M Bismark and R Paterson “No-fault compensation in New Zealand:
Harmonizing injury compensation, provider accountability,
safety” (2006) 25(1) Health Affairs 278.
36 A Kachalia and others “Beyond negligence: Avoidability and medical injury compensation”
slow, expensive and adversarial focus on finding fault.37 The process of lodging a complaint or making a notification is simpler and less costly than filing a law suit. Complaints commissioners can offer a range of non-monetary remedies as part of their complaint resolution processes. Mandatory reporting requirements oblige practitioners and employers to report concerns about impaired colleagues to regulators.38 And, in New Zealand, no-fault compensation is available to injured patients without the requirement to demonstrate negligence.39 Together these features fill many of the gaps long lamented in the medical negligence system.40
And yet, even these medico-legal systems fall short in meeting the expectations of patients, practitioners, and the public. Indeed, some of the discourse surrounding medico-legal agencies suggests that we may have achieved “the worst of both worlds”:41 a system of regulation that goes too far on one hand, and on the other hand does too little.
2.7.7 Regulation goes too far
Medical practitioners and their professional bodies have argued that the actions of medico-legal agencies lead, at various times, to “futility, perversity, and jeopardy” (to borrow from the words of Albert Hirschman).42 The argument of futility rests on the premise that the healthcare system and medical culture are complex systems and that the simplistic levers exercised by medico-legal agencies do little to alter the underlying status quo. In practice, the values and norms that are intrinsic to the practice of medicine will often preside over “clumsy and unsatisfactory” external controls.43
(2008) 66(2) Social Science & Medicine 387; Peter Schuck “Tort reform, Kiwi-style” (2008) 27 Yale Law & Policy Review 187.
37 M Mello, D Studdert and T Brennan “The new medical malpractice crisis” (2003) 348 New England Journal of Medicine 2281; D Skegg “Oral contraceptives, venous thromboembolism, and the courts: There has to be a better way of resolving claims about adverse effects” (2002) 325(7363) British Journal of Medicine 504.
38 M Bismark and others “Mandatory reports of concerns about the health, performance and conduct of health practitioners” (2014) 201(7) The Medical Journal of Australia 399.
39 M Bismark and R Paterson “No-fault compensation in New Zealand: Harmonizing injury compensation, provider accountability, and patient safety” (2006) 25(1) Health Affairs 278.
40 M Mello, D Studdert and T Brennan “The new medical malpractice
crisis” (2003) 348 New England Journal of Medicine 2281; A Kachalia and Mi
Mello “New directions in medical liability reform” (2011) 364(16)
New England Journal of Medicine
41 T Stoppard Rosencrantz and Guildenstern are dead (Faber, London, 1967) at 52.
42 A Hirschman The rhetoric of reaction: Perversity, futility, jeopardy (Harvard University Press, Cambridge 1968).
43 C Paul “Internal and external morality of medicine: Lessons from New Zealand” (2000) 320(7233) British Medical Journal 499 at 503.
The argument of perversity suggests that, in some cases, medico-legal action will produce an outcome opposite to that which was intended. For example, practitioners commonly voice concerns that mandatory reporting of impaired practitioners to regulators may drive those practitioners with health concerns further underground.44 In another example, negligence claims may lead to defensive medicine, whereby practitioners order unnecessary tests or decline to see high-risk patients, resulting in poorer care than would otherwise be provided.45
And finally, the argument of jeopardy claims that while medico-legal actions may have some beneficial consequences, they also result in unintended harms in other parts of the health sector. For example, some practitioners have argued that revalidation takes time and resources away from more pressing priorities in the health sector.46
2.1.2 Regulation does too little
In contrast, many patients, families and their advocates believe that current systems of medico-legal regulation do not go far enough. They feel frustrated and unheard by medico-legal processes that are difficult to access, slow to act, and impersonal in their response.47 And all too often, those processes seem to ‘miss the point’ of the patient’s claim or complaint. It is not unusual for a complainant to lodge a claim or complaint, only to conclude: “this isn’t the type of process I was looking for”.48
Patients and families also resent the ‘wall of silence’ that seems to exist around many medico-legal processes. We live in a world where information on the most esoteric of subjects is available at the touch of a screen. Consumers are able to access real-time information on the quality and safety of travel destinations, chefs, airline pilots, and funeral directors. Yet, when it comes to choosing a doctor or following the progress of a notification to a medical board, medico-legal agencies offer little meaningful information. Indeed, the identities of those who have been
44 R Beran “Mandatory notification of impaired doctors” (2014) 44(12a) Internal Medicine Journal 1161.
45 W Cunningham and S Dovey “Defensive changes in medical practice and the complaints process: A qualitative study of New Zealand doctors” (2006) 119(1244) Journal of the New Zealand Medical Association 2283.
46 S Wilson “Revalidation: safe practice or just another massive hurdle?” (2013) 53(3) Medicus 20.
47 Health Issues Centre Setting things right: Improving the consumer experience of AHPRA (Melbourne, June 2014).
48 Health Issues Centre Setting things right: Improving the consumer experience of AHPRA (Melbourne, June 2014) at 7.
subject to complaints and disciplinary proceedings are often shielded behind opaque processes and orders for name suppression.49
2.2 Shared Concerns About Regulatory Effectiveness
On their surface, the criticisms expressed by patients and practitioners take quite different forms. However, deeper analysis suggests that they are connected by three broader concerns. These are: a case-by-case focus that sometimes misses the bigger picture; a reactive rather than proactive approach; and a process-driven system that, at times, values ‘doing it the right way’ over ‘doing the right thing’.
2.2.1 Case-by-case focus
The first common concern is that the ‘visual field’ of individual staff working within medico-legal agencies is largely restricted to the individual cases that come before them. This is a problem for two reasons. First, without a coherent and principled understanding of how an individual case fits into a broader decision-making framework, decisions may be disproportionate or inconsistent. Under current approaches, medico-legal agencies are blinkered in their ability to see how a complaint fits in with a particular practitioner’s history, wider feedback from other patients and peers, and information held by other agencies. This is of concern to both practitioners and patients as it can result in two kinds of errors: overlooking legitimate risks or taking regulatory action against practitioners who do not pose a risk.
Secondly—and of more concern to the broader public—processing cases oneby-one does not encourage, nor leave much time for, higher-level thinking and analysis. Therefore, opportunities to develop systemic and lasting solutions are lost.50 Recommendations for improvement tend to occur in an incremental and piecemeal fashion, lacking a sense of wider vision. In essence, medico-legal agencies offer reactive solutions to immediate problems, while paying little attention the systemic problems underlying them.
Unfortunately, even if medico-legal agencies do wish to see a bigger picture, the necessary information is often difficult to access and gather.51 Insurers,
49 M Bismark and others “Claiming behaviour in a no-fault system of medical injury: A descriptive analysis of claimants and non-claimants” (2006) 185(4) Medical Journal of Australia 203.
50 M Sparrow The regulatory craft: Controlling risks, solving problems, and managing compliance (Brookings Institution Press, Washington, 2000).
51 M Kleiner Stages of occupational regulation: Analysis of case studies (Upjohn Institute Press, Kalamazoo, 2013).
commissioners, and regulatory boards typically only hear the voices of patients and practitioners whose cases are brought before them. Yet, international research suggests that under-reporting of adverse events is rife.52 Without such denominator data, it is hard for any agency to even recognise its own blind spots. Where data is available, agencies may not have the skills or knowledge to make maximal use of it.53 While the rest of the world moves rapidly into the world of “big data”,54 medical regulatory culture lags behind in realising the benefits of large-scale analysis.
2.2.2 Reactive approach
The second common concern relates to the reactive nature of most insurers, commissioners, and regulatory boards. In the wider world of regulation, academics have argued since at least the 1970s that merely reacting to individual cases as they present themselves is an inefficient and ineffective use of resources. Instead, agencies should learn to proactively identify underlying problems and tailor-make solutions that prevent recurrence.55 The same is likely to be true in healthcare. For example, it is well-established that malpractice lawsuits,56 complaints,57 and disciplinary proceedings58 tend to cluster among relatively small groups of doctors. Yet, regulators know little about the characteristics of such practitioners beyond anecdote and intuition.
Consequently, the medico-legal sector remains a reactive one, dealing primarily with the aftermath of adverse events and behaviours.59 For patients, this is problematic because they remain at risk of harm until appropriate action is taken. For practitioners, it is problematic because, without proactive support and early
52 A Öhrn and others “Reporting of sentinel events in Swedish hospitals: A comparison of severe adverse events reported by patients and providers” (2011) 37(11) Joint Commission Journal on Quality and Patient Safety 495.
53 L Leape and J Fromson “Problem doctors: Is there a system-level solution?” (2006) 144(2) Annals of Internal Medicine 107.
54 J Manyika and others Big data: The next frontier for innovation, competition, and productivity (McKinsey Global Institute, May 2011); E Brynjolfsson and A McAfree “Big data: The management revolution” (2012) 90(10) Harvard Business Review 61.
55 H Goldstein “Improving policing: A problem-oriented approach” (1979) 25(2) Crime & Delinquency 236.
56 F Sloan and others “Medical malpractice experience of physicians:
Predictable or haphazard?” (1989) 262(23) JAMA 3291.
57 G Hickson “Patient complaints and malpractice risk” (2002) 287(22) JAMA 2951.
58 J Morrison and P Wickersham “Physicians disciplined by a state medical board” (1998) 279(23) JAMA 1889.
59 M Bismark “Identification of doctors at risk of recurrent complaints: A national study of healthcare complaints in Australia” (2013) 22 BMJ Quality & Safety 532.
intervention, minor health and competence issues can evolve into full-blown impairments, jeopardising their careers and wellbeing.
Regulatory expert Professor Malcolm Sparrow urges regulators to “pick important problems and fix them”.60 And indeed, many agencies would like to take a more proactive approach. In the words of former New Zealand Health and Disability Commissioner Ron Paterson, health complaints commissioners would prefer to be the fence at the top of the cliff, rather than the ambulance at the bottom.61 However, to date, agencies have lacked the tools to proactively identify high-risk groups of practitioners and intervene early to eliminate or reduce risk. Instead, they continue to live with the fear that one day a patient will die after red flags were overlooked and someone will ask: why didn’t you act sooner?
2.2.3 Process-driven systems
The final common concern is the process-driven nature of medico-legal agencies. In practice, the time and efforts of medico-legal agencies are overwhelmingly focused on receiving, prioritising, and handling cases in a procedurally ‘correct’ manner. This in turn means that communications are shaped by workflows and precedents, rather than human relationships or outcomes. And remedies are largely determined by what the agency routinely offers, rather than the resolution that would be most effective at protecting patients, while supporting practitioners to practice safely.
The consequence is a process that is stressful and uncertain for both patients and practitioners.62 A recent review of AHPRA processes found that people dealing with the agency wanted to deal consistently with the same staff, be communicated with in a personal way, and to know and experience that the agency understood them, heard them, responded by taking them seriously, kept them informed, and explained the reasons for its actions.63
60 M Sparrow The regulatory craft: Controlling risks, solving problems,
and managing compliance (Brookings Institution Press, Washington, 2000) at
61 Ron Paterson “Farewell Speech” (speech to Parliament, Wellington, 2010).
62 W Cunningham and S Dovey “Defensive changes in medical practice and the complaints process: A qualitative study of New Zealand doctors” (2006) 119(1244) Journal of the New Zealand Medical Association 2283; L Nash, C Tennant and M Walton “The psychological impact of complaints and negligence suits on doctors” (2004) 12(3) Australasian Psychiatry 278.
63 Health Issues Centre Setting things right: Improving the consumer experience of AHPRA (Melbourne, June 2014) at 48.
Both patients and practitioners would benefit from a more outcome-focused approach with less reliance on pro-forma responses and more attention to the voices of patients and practitioners.64 However, in order to achieve resolutions that better meet patients’ needs, agencies first need a clear understanding of what those needs are and where existing gaps eventuate. It is in developing this understanding that a new, epidemiology-inspired approach shows great promise.
3. AN EPIDEMIOLOGICAL APPROACH
Traditionally, medico-legal agencies have focused on the resolution of individual cases. When they have applied analytical tools, these were often drawn from legal scholarship with a focus on legal principles and precedent. While such tools have their place, they offer a narrow visual field, constraining the ability of agencies to see the patterns that are writ large across hundreds or thousands of cases.
Understanding the needs of injured patients and the limitations of current medico-legal systems requires lifting our gaze to look beyond individual cases to identify patterns and compare the characteristics of different groups of patients and practitioners. The medico-legal process is not designed for such analyses. But the science of epidemiology is.65
Epidemiology is commonly defined as the branch of medicine which deals with the incidence, distribution, and possible control of diseases and other factors relating to health. It spans the spectrum from confirming the existence of a problem, to understanding causal factors, to implementing and evaluating preventive measures.
Three broad methodological approaches used by epidemiologists are: descriptive (organising information by person, place and time), analytic (e.g. incorporating a case-control or cohort study), and experimental (e.g. randomly allocating individuals to receive an intervention or a placebo).
64 Health Issues Centre Setting things right: Improving the consumer experience of AHPRA (Melbourne, June 2014).
65 K Rothman, S Greenland and T Lash Modern epidemiology (Lippincott Williams & Wilkins, Philadelphia, 2008).
3.1.1 Descriptive approaches
Descriptive statistics are the most familiar and straightforward form of analysis. These are the types of counts and percentages typically found in the annual reports published by medico-legal agencies each year. A weakness of many descriptive statistics is the use of a floating numerator, where the number of cases is not related to an appropriate ‘at risk’ population. For example, most complaints against doctors involve general practitioners. However, this tells us little about the relative complaint risk of general practitioners compared with other doctors because general practitioners account for the bulk of the medical workforce. Once one accounts for the number of doctors in each speciality (using data from the medical register), it becomes apparent that surgeons are actually at significantly higher risk of complaints than general practitioners.
3.1.2 Analytic approaches
Analytic studies adopt a more sophisticated approach, and—in the medico-legal setting—have proven to be particularly useful for identifying risk factors for claims or complaints. Three analytic approaches are cross-sectional studies, case-control studies, and cohort studies.
In a cross-sectional study, a population is surveyed at a particular point in time. For example, patients who lodged a complaint could be asked about their motivation for taking medico-legal action, or nurses could be surveyed about the impact of a complaint on their professional practice.
The case-control method starts with a case group of individuals with a particular condition and compares them with a similar group of individuals without the condition. An example of a case-control study is selecting doctors with a prescription drug addiction, and a comparison group who do not have a drug addiction (matched for age, sex, and specialty), to see what factors may have influenced occurrence in the group that developed the problem.
A cohort study observes a population—some of whom were exposed to a potential causal factor and some of whom were not—to see who develops the outcome of interest. An example of a cohort study is an analysis of two groups of doctors, one composed of doctors who trained in their current country of practice, and the other of doctors who trained elsewhere. The occurrence of disciplinary actions in both groups allows you to make an association between country of training and risk of disciplinary action.
More sophisticated statistical methods such as time-to-event method of analysis enable us to determine the characteristics of doctors likely to incur recurrent
complaints and to estimate each practitioner’s risk of complaint recurrence at specific time points.
3.1.3 Experimental approaches
The third ‘gold standard’ epidemiologic method is the experimental approach. As its name suggests, experimental methods involve manipulating certain conditions in order to test a hypothesis. For example, a medical board could randomly allocate doctors with a history of bullying and harassment to either a series of intensive communication skills workshops or to workplace supervision, to assess which (if either) had a greater impact on their workplace conduct. This conduct might be assessed, for example, via 360-degree surveys of patients, managers, and colleagues. While experimental designs are common before introducing new pharmaceuticals, and indeed often required, they are rarely employed before imposing new regulatory initiatives such as revalidation or mandatory reporting. This is despite the fact that the costs, benefits, and harms to public health may be even larger than for a new medication.
Unfortunately, epidemiological research in the field of medical regulation and healthcare improvement is not without limitations. When undertaking such research with medico-legal agencies, common barriers include poor data quality, inconsistent coding taxonomies, and a paucity of robust denominator data.
3.2.1 Data collection
Medico-legal data in Australia and New Zealand, where my group has undertaken much of our work, remains widely dispersed across different agencies including indemnity insurers, complaints commissioners, the Accident Compensation Corporation, and health practitioner regulation boards.66 The establishment of the Australian Health Practitioner Regulation Agency has helped to address this problem by bringing together over seventy health practitioner regulation boards under one umbrella. However, data relating to complaints and claims continues to be spread across multiple health complaints commissioners, insurers, coroners, employers, and even law enforcement agencies. As a result, data collection and data linkage between agencies remains a time-consuming and labour-intensive process. This problem is not unique to Australia and New Zealand.
66 M Walton and others “Health complaint commissions in Australia: Time for a national approach to data collection” (2012) 11(1) Australian Review of Public Affairs 1.
3.2.2 Data completeness and coding
In many contexts, the ability to analyse crucial questions is also hampered by the lack of principled and consistent taxonomies for coding important variables.67 For example, when working with data from complaints commissioners, our group has been unable to code data according to the merit of complaints. This is because complaints commissioners focus on resolving complaints, rather than issuing decisions as to whether those complaints were substantive or upheld as valid. In another example, coding of complaints often fails to identify the actual harm or outcomes associated with the issues identified. Coding the broad issues identified in a complaint offers some clues to the possible severity of harm. However, it is at best a crude proxy.
Completeness and accuracy of data are also notable concerns. For example, in Australia, some key variables such as ethnicity are not collected or are only collected by some agencies. Overcoming this limitation will be particularly important for understanding access to medico-legal agencies by indigenous Australians, a group who already experience a cascade of disparities within the healthcare system. The extent of problems relating to data completeness varies widely and is slowly improving.
The need for international standardised terminology, common methods of measurement and evaluation, and compatible reporting of adverse events has been recognised for at least a decade.68 However, these goals remain elusive within the international patient safety community.
3.2.3 Denominator data
Finally, the problem of finding robust denominator data is also far from solved. In particular, adverse event data remains hard to come by. This is largely due to the enormous time and cost associated with undertaking the large-scale file reviews required to identify all adverse events within a patient population. Obtaining workforce denominators is easier, although medical registers typically only provide head counts of practitioners, rather than more sophisticated measures of doctors’ exposure to complaint risk, such as full-time equivalents, patient volumes, or case complexity.
67 K Elkin and others “Removal of doctors from practice for professional misconduct in Australia and New Zealand” (2012) 21(12) BMJ Quality & Safety 1027.
68 J Loeb and A Chang Patient safety: Reduction of adverse events through common understanding and common reporting tools. Towards an international patient safety taxonomy: A review of the literature on existing classification schemes for adverse events and near misses; a draft framework to analyze patient safety classifications (World Health Organisation, Geneva, 30 June 2003).
3.3 A Social Movement, a Scientific Awakening
Despite these limitations, researchers around the world have begun to identify and harness the potential of epidemiological approaches to illuminate and inform medical regulation, healthcare safety, and the medico-legal sphere. At the same time, the patient safety movement is gaining momentum, driven by individual and institutional advocates for change. This movement has begun to promote the importance of previously under-utilised evidence-based approaches to healthcare regulation and improvement. It is within this context, at the intersection of a social movement and a scientific one, that researchers have found fertile ground for exploration, and the potential to make a real-world difference to the safety and quality of healthcare.
4. FINDINGS AND OPPORTUNITIES FOR RESEARCH TRANSLATION
Over the last fifteen years, my research team—led by Professor David Studdert has applied epidemiological approaches to a range of medico-legal questions. Sometimes, our research confirmed an intuition. Sometimes it dispelled a myth. And sometimes, it uncovered a new and surprising finding that shifted our thinking in a new direction.
4.1 Summary of Findings 4.7.7 Who complains?
Our team’s early exploration of epidemiological approaches to medico-legal problems was inspired by early work in the US and elsewhere that highlighted the alarming frequency of preventable harm in care. These figures raised questions about the degree to which rates of harm were reflected in patient complaints and compensation claims. To examine this relationship, we sought to estimate the proportion and characteristics of patients injured by medical care in New Zealand public hospitals who went on to complain to the Health and Disability Commissioner, or seek compensation through the Accident Compensation Corporation.
The percentage of injured patients who lodged claims or complaints was estimated by linking the Commissioner’s complaints database to records reviewed in the New Zealand Quality of Healthcare Study (NZQHS). Bivariate and multivariate analyses investigated sociodemographic and socioeconomic differences between those who claimed or complained, and those who did not.
We found that, following an adverse event, only a small proportion of
patients will bring a claim or complaint. Among serious, preventable
events identified by the New Zealand Quality of Health Study, only 4% resulted
Among those patients who were eligible for no-fault compensation, fewer than 5% claimed it.70
At a population level, there were significant differences between those who took medico-legal action and those who did not. As expected, the propensity of injured patients who complained increased with the severity of injury. Odds of complaint were more than ten times greater for serious and permanent injuries as opposed to temporary ones.71 Of more concern was our finding that the odds of complaining were significantly lower for patients who were elderly, of Māori ethnicity, or lived in the most deprived areas.72 Previous studies in the United States provide similar findings.73 However, we were surprised to find that such profound under-claiming and access disparities persisted in New Zealand, where no-fault compensation and complaints processes were intended to be accessible to all.
4.1.2 Motivations and expectations of complainants
Our analysis of claims and complaints showed that patients who do pursue medico-legal action in the aftermath of medical injury seek four forms of accountability:74
69 M Bismark and others “Relationship between complaints and quality of care in New Zealand: A descriptive analysis of complainants and non-complainants following adverse events” (2006) 15(1) Quality and Safety in Health Care 17.
70 M Bismark and others “Claiming behaviour in a no-fault system of medical injury: A descriptive analysis of claimants and non-claimants” (2006) 185(4) Medical Journal of Australia 203.
71 M Bismark and others “Relationship between complaints and quality of care in New Zealand: A descriptive analysis of complainants and non-complainants following adverse events” (2006) 15(1) Quality and Safety in Health Care 17.
72 M Bismark and others “Relationship between complaints and quality of care in New Zealand: A descriptive analysis of complainants and non-complainants following adverse events” (2006) 15(1) Quality and Safety in Health Care 17.
73 P Raj and G Deka (eds) Handbook of research on cloud infrastructures for big data analytics (Information Science Reference, United States of America, 2014).
74 M Bismark and E Dauer “Motivations for medico-legal action: Lessons from New Zealand” (2006) 27(1) The Journal of Legal Medicine 55.
Among patients in New Zealand who took medico-legal action, 50% sought corrective action to prevent similar harm to future patients. Typical comments included: “I hope that this complaint makes a difference for the treatment of others” and “We certainly wouldn’t want anyone else to go through what we went through.” Forty percent wanted better communication with themselves and/or their family.75
Injured patients and their families involved in medico-legal action have many objectives other than monetary ones. However, for some patients, apologies and explanations do not suffice as a substitute for financial compensation. The odds that patients would seek compensation were significantly increased if they were in their prime working years or had a permanent disability as a result of their injury.76
Unfortunately, there is evidence of a significant ‘expectation gap’ between what patients seek and what they eventually get out of the complaints process. Analysing a sample of complaints relating to informed consent in Victoria, Australia, we found that just one third of complainants who sought restoration received it. And only one in five complainants who sought correction received assurances that changes had been made to prevent future harm. There are two logical ways in which the expectation gap can be narrowed: either by decreasing unrealistic expectations or providing more complainants with the remedies they seek.77
4.1.3 Clinical risk factors for complaints
Just as certain practitioners appear to be at higher risk of complaint, so too are certain types of procedures and complications. Among 481 malpractice claims and serious healthcare complaints involving informed consent, we found that a disproportionate share (16%) related to cosmetic procedures. Based on a detailed review of these cases, it seems that certain factors ‘supercharge’ a healthcare interaction for a medico-legal complaint: an elective procedure, a rushed and pressured consultation, an unrealistic portrayal of benefits, and highly visible complications.78 In a separate analysis—co-authored with Atul Gawande—we
75 M Bismark and E Dauer “Motivations for medico-legal action: Lessons from New Zealand” (2006) 27(1) The Journal of Legal Medicine 55.
76 M Bismark and others “Accountability sought by patients following adverse events from medical care: The New Zealand experience” (2006) 175(8) Canadian Medical Association Journal 889.
77 M Bismark and others “Accountability sought by patients following adverse events from medical care: The New Zealand experience” (2006) 175(8) Canadian Medical Association Journal 889.
78 M Bismark and others “Legal disputes over informed consent for cosmetic procedures: A descriptive study of negligence claims and complaints in Australia” (2012) 65(11) Journal of
identified a small group of adverse outcomes that seem to matter a lot to patients, yet may not be routinely disclosed during the informed consent process. These were poor cosmetic result, the need for further surgery, impaired vision or hearing, chronic pain, and infertility or sexual dysfunction.79
4.1.4 Distribution of complaints among practitioners
While medical practitioners feel that they all practice under a medico-legal cloud, our research confirmed that it “doesn’t rain on everybody equally and that some physicians have a malpractice dark cloud”.80 Among a national sample of nearly 19,000 complaints to health complaints commissioners, we found that just 3% of all doctors accounted for nearly half of all complaints.81 While we expected to find some clustering—based on previous studies of malpractice claims, complaints, and disciplinary actions—we were surprised by the extent to which some doctors were over-represented in complaints data.
The number of prior complaints doctors had experienced was a strong predictor of subsequent events, and a dose-response relationship was evident.82 Doctors named in a third complaint had a nearly 60% probability of being named in a further complaint within two years.83 Demographic factors also played a part. In particular, male doctors had a 40% higher risk of recurrence than their female colleagues.84
Applying similar methodology to malpractice claims data from the United States resulted in similar findings. Using data from the National Practitioner Data Bank, we analysed 66,426 claims paid against 54,099 physicians over a ten year period (2005 through 2014). We calculated concentrations of claims among physicians and used survival analysis to identify characteristics of those at high risk of recurrent claims. As in our Australian study, a small number of physicians with distinctive
Plastic, Reconstructive & Aesthetic Surgery 1506.
79 M Bismark and others “Legal disputes over duties to disclose treatment risks to patients: A review of negligence claims and complaints in Australia” (2012) 9(8) PLOS Medicine, online edition, e1001283.
80 Interview with Gerald Hickson (R Wachter, Perspectives on Safety, 1 December 2009) transcript provided by AHRQ Patient Safety Network (United States of America).
81 M Bismark and others “Identification of doctors at risk of recurrent complaints: A national study of healthcare complaints in Australia” (2013) 22 BMJ Quality & Safety 532.
82 M Bismark and others “Identification of doctors at risk of recurrent complaints: A national study of healthcare complaints in Australia” (2013) 22 BMJ Quality & Safety 532.
83 M Bismark and others “Identification of doctors at risk of recurrent complaints: A national study of healthcare complaints in Australia” (2013) 22 BMJ Quality & Safety 532.
84 M Bismark and others “Identification of doctors at risk of recurrent complaints: A national study of healthcare complaints in Australia” (2013) 22 BMJ Quality & Safety 532.
characteristics accounted for a disproportionately large number of claims. Just 1% of all physicians accounted for 32% of paid malpractice claims. Again, past behaviour was the strongest predictor of future behaviour. For example, as compared with physicians who had one previous paid claim, the 2,160 physicians who had three paid claims had three times the risk of incurring another.85
4.1.5 Mandatory reporting
Mandatory reporting is the legal duty of practitioners to report serious concerns about the health, conduct, or performance of other practitioners to the regulator. It offers another potential way for regulators to identify high-risk practitioners. However, early data from Australia’s new mandatory reporting regime suggests that many practitioners remain unsure about the nature and scope of their obligations.86 We carried out a retrospective review and analysis of more than 800 instances in which issues were reported by practitioners to the Australian Health Practitioner Regulation Agency, under the auspices of mandatory reporting, between November 2011 and December 2012.87 While the overall incidence of such reports was very low, we noted wide variation in reporting rates between jurisdictions and professions. While the mandatory reporting laws apply to all fourteen registered health professions in Australia, nearly 90% of notifications involved a doctor or nurse in the role of notifier and/or respondent. Male practitioners were two and a half times more likely to be subject to a mandatory report than female practitioners.
Of particular concern was the low rate of notifications by nurses about doctors (3%). Although nurses are often well placed to observe poorly performing doctors, our data suggests that the new law has not overcome previously identified factors that may make it difficult for nurses to report concerns about doctors.88
We also observed important differences between the statutory threshold for reporting and the factors that practitioners actually weighed in their decision-making. For example, reporters often referred to an impaired practitioner’s lack of
85 D Studdert and others “Prevalence and characteristics of physicians prone to malpractice claims” (2016) 374(4) New England Journal of Medicine 354.
86 M Bismark and others “Mandatory reports of concerns about the health, performance and conduct of health practitioners” (2014) 201(7) The Medical Journal of Australia 399.
87 M Bismark and others “Mandatory reports of concerns about the health, performance and conduct of health practitioners” (2014) 201(7) The Medical Journal of Australia 399; M Bismark “Reporting of health practitioners by their treating practitioner under Australia’s national mandatory reporting law” (2016) 204(1) The Medical Journal of Australia 24.
88 M Bismark and others “Mandatory reports of concerns about the health, performance and conduct of health practitioners” (2014) 201(7) The Medical Journal of Australia 399.
honesty, insight, or compliance with treatment as an influential factor in deciding to make a report. Yet these factors are not explicitly mentioned in the mandatory reporting legislation.
Our research has identified key challenges for mandatory reporting, including ensuring the threshold for reporting is appropriately defined and clearly understood, improving access to evidence-based health programs, and strengthening upstream protections to prevent impairment at its roots.89
4.2 Recurring Themes
By applying an epidemiological lens to medico-legal data, three powerful and recurring themes have emerged. These themes provide insights that extend beyond the specifics of the above-summarised data to shine a light on safety and harm in the healthcare system, and ways in which the balance between the two may be redressed.
4.2.1 Unheard versus valued voices
First, patients and practitioners serve an important surveillance role and can be powerful allies in the quest for safer care. Their eyes and ears are effective in identifying problems, but only if medico-legal agencies are committed to listening, and making it safe and worthwhile for patients and practitioners to speak up. Hearing the voices of patients is also crucial for providing them with the remedies that they want and need when confronted with harm in healthcare.
In practice, the voices of many injured patients go unheard, either because they never file a claim or complaint, or because there is a mismatch between what they seek, and what an entity is equipped to deliver. In particular, the relatively low propensity to complain among patients who are elderly, socio-economically deprived, or of Māori ethnicity suggests troubling disparities in access to, and utilisation of, claims and complaints procedures.90 In thinking about these muffled voices, we would do well to remember the words of Arundhati Roy: “There’s really
89 M Bismark, J Morris and C Clarke “Mandatory reporting of impaired medical practitioners: Protecting patients, supporting practitioners” (2014) 44(12a) Internal Medicine Journal 1165.
90 M Bismark and others “Claiming behaviour in a no-fault system of medical injury: A descriptive analysis of claimants and non-claimants” (2006) 185(4) Medical Journal of Australia 203.
no such thing as the ‘voiceless’. There are only the deliberately silenced or the preferably unheard.”91
4.2.2 Blinkered versus broad focus
Secondly, important patterns have been overlooked because agencies have lacked the time, resources, and analytical skills to make sense of population-level data. The claims and complaints we reviewed are a rich and nuanced source of data on risks to patient safety and sources of patient dissatisfaction. But once an individual file has been closed, it is usually archived without further analyses of how it contributes to broader patterns of concern. Agencies will commonly undertake some descriptive analyses for their annual reports, but these tend to focus on numbers of cases resolved and days to closure, rather than contributing to an improved understanding of risks and causal factors. A better understanding of ‘hot spots’ of medico-legal concern has real potential to inform the way we train health practitioners, communicate with patients, and design regulatory systems.
4.2.3 Reactive versus proactive regulation
And thirdly, serious adverse events rarely happen without warning. Our research shows a significant concentration of risk among a relatively small number of health practitioners. Much like detection dogs are used to sniff out security risks from among the thousands of passengers who pass through an airport each day, powerful statistical methods have the potential to help medico-legal agencies identify those few practitioners who are at significantly increased risk of future complaints compared with their peers.92 Having identified the concentration of risk, the next challenge for regulators is to develop and evaluate programs to try and reduce that risk.
Collectively, our findings suggest that a reactive and process-driven approach to resolving individual cases has ‘dulled the senses’ of medico-legal agencies. It is not that they do not want to hear patients’ and practitioners’ voices at the population level or see patterns of concern. Rather, as organisations, they do not have the right skills and experience to do so. By aggregating information and comparing groups, as epidemiologists do, medico-legal agencies are more likely to be able to see the patterns that will support evidence-based decisions. Perhaps it is time for
91 A Roy “Peace, The New Corporate Liberation Theology” (Sydney Peace Prize Lecture, Sydney, 4 November, 2004).
92 M Bismark, M Spittal and D Studdert “Prevalence and characteristics of complaint-prone doctors in private practice in Victoria” (2011) 195(6) Medical Journal of Australia 25.
the medical and legal professions to recognise that epidemiology is as indispensable to medical regulation as it is to public health.
4.3 Opportunities for Research Translation
There is considerable need and scope for further research, anchored in epidemiological approaches, to further our work in healthcare regulation, quality and safety. For example, research is needed to create more sophisticated methods for risk prediction as it applies to practitioners’ potential to cause harm. Also, we must go beyond merely identifying risks, and at-risk individuals, to designing preventative and restorative interventions. Furthermore, research into the safety impacts of forces and entities arising beyond the walls of regulators—including profession and institutional cultures within healthcare and the role of social inequities and health disparities—is essential.
However, our findings, as summarised above, already point to some practical actions, approaches, and changes that may translate our growing knowledge into practical improvement. With due acknowledgment of the research that remains to be done, we would be remiss not to examine the opportunities to translate lessons from our existing work into interventions that may improve healthcare safety and regulatory effectiveness.
4.3.1 Capitalising on regulatory data
Regulators and legislators should invest in and prioritise the skills, technology, policies, and inter-agency cooperation required to capitalise on the wealth of data that becomes available to them in the course of their operation. Such a forward-thinking approach is likely to bring benefits that increase in magnitude, scope, and potency over time.
Coming to view the world as consisting of populations and patterns—not just individuals and chance—will be an essential element of effective, modern, evidence-based regulation. One that emphasises prevention over reaction and early intervention over after-harm sanctions. And one that achieves ‘right touch regulation’: the ability to make informed and responsive judgments about when to apply lighter approaches, versus more-heavy-handed options.93
93 H Clayton and K Webb “The benefits of a ‘right-touch’approach to health care regulation” (2014) 19(4) Journal of Health Services Research & Policy 198; J Healy and J Braithwaite “Designing safer health care through responsive regulation” (2006) 184(10) Medical Journal of Australia S56.
4.3.2 Diversifying input
Regulators and legislators also need to carefully consider ways to capture, amplify, and do justice to the voices of vulnerable patients and groups. This means groups who may not have the cultural capital and confidence to confide in a medico-legal agency. Low rates of complaints lodgement and reports from harmed patients are concerning in their own right. Even lower rates among vulnerable groups of patients, a pattern which may extend to other potential whistle-blowers, represents both a betrayal of social justice, and a large and deeply concerning loss to the healthcare system’s potential for improvement.
4.3.3 Addressing cultures and norms
All who participate in, endorse, benefit from or influence the culture, beliefs and norms of the health professions must maintain a critical awareness of the impact these have on the safety and quality of care. With respect to regulators, they are more likely to succeed if they use strategies that are responsive to the cultural norms and beliefs of those being regulated.94
With respect to practitioners, re-examining healthcare culture and norms includes reconsidering the way complaints are viewed. Most helpfully, this will involve seeing them as an invitation to learn and improve, rather than something to be feared, dismissed or resented.95 External regulation will be of little use in establishing and maintaining safer healthcare without a functioning internal morality within individual practitioners, professions and healthcare services.96 Crucial to this transition is recognising a tenet of prevention alluded to by our findings to date—that both patients and practitioners will benefit when patterns of concern are identified and addressed sooner rather than later.
Critical consideration of healthcare culture also includes accepting that being part of a profession carries a responsibility to help ensure that professional peers are “good enough”.97 Ideally, the profession and its constituent members should have the courage, integrity and moral fortitude to be able to identify and provide
94 H Clayton and K Webb “The benefits of a ‘right-touch’ approach to health care regulation” (2014) 19(4) Journal of Health Services Research & Policy 198; J Braithwaite Restorative justice & responsive regulation (Oxford University Press, New York, 2002).
95 A Kachalia “Improving patient safety through transparency” (2013) 369(18) New England Journal of Medicine 1677.
96 C Paul “Internal and external morality of medicine: Lessons from New Zealand.” (2000) 320(7233) British Medical Journal 499.
97 R Paterson The good doctor: What patients want (Auckland University Press, Auckland, 2012) at 20.
support to colleagues at risk—from medical school through to retirement—long before medico-legal action is required.98
Twenty-five years on from the landmark Cartwright Inquiry in New Zealand, the medico-legal systems of Australia and New Zealand are widely regarded as a promising alternative to the pitfalls of medical malpractice litigation.99 However, despite their advantages over a court-based system, concerns remain: regulatory systems can be daunting to access, slow to act, and sometimes seem to ‘miss the point’ with their reactive and process-driven approach.
In addressing these problems, regulators are constantly faced with choices about how to allocate scarce resources. Epidemiological approaches offer one way for medical-legal agencies to make these choices more rationally and analytically. Population thinking is a skill that has to be acquired. It does not emerge naturally from common life experience. Yet many regulators are short on the requisite people, skills, and analytical versatility to undertake the kinds of analyses needed to make sense of the data they collect at a population level.100
We do not suggest that all medico-legal decisions should be evidence-based. Just as evidence-based medicine integrates an individual’s skills and assessment with the best available evidence from medical research, so too should evidence-based policy. As expressed by Shojania and others, insistence on evidence should not “prevent implementation of practical, low-risk, but understudied interventions that seem likely to work.”101 For some regulatory decisions, there are no objective criteria with which one can locate the optimal balance between competing interests. These decisions can be informed, but not resolved, by epidemiology. Our hope is that an epidemiological approach to medico-legal data—an evolving stream of research at the confluence of Professors Peter and David Skegg’s scholarship will complement and enhance medico-legal agencies’ existing efforts to support practitioners and protect patients from harm.
98 M Papadakis “Unprofessional behavior in medical school is associated with subsequent disciplinary action by a state medical board” (2004) 79(3) Academic Medicine 244.
99 PDG Skegg “Compensation in the New Zealand health care sector” in J Dute, M Faure and H Koziol (eds) No-fault compensation in the health care sector (Springer, Vienna, 2004) 283.
100 M Sparrow The regulatory craft: Controlling risks, solving problems, and managing compliance (Brookings Institution Press, Washington, 2000).
101 K Shojania and others “Safe but sound: Patient safety meets evidence-based medicine” (2002) 288(4) JAMA 508 at 512.