Preface to the first edition

In looking for a way to introduce this book, I came across a speech by Admiral Hyman G. Rickover, delivered at Columbia University in 1982, in which he said,

Several citations of this quote on the internet present a curious discrepancy, in that the word “impatience” is replaced with “patience”. I find this variation to be intriguing, as it seems to me that both courageous patience and courageous impatience must be manifest for the greatest progress to occur. If one has a good idea, one must not only be both steadfast and patient to stay the course and see it through, but also, at the same time, one must continue to push aggressively, even impatiently, lest the effort lose momentum and flag.

My idea for this book was based on the observation that computer-based clinical decision support (CDS) has the potential to be truly transformative in health care but that, despite considerable creativity and experimentation by enthusiasts over more than four decades, and convincing demonstration of effectiveness in particular settings, the adoption of CDS has proceeded at a snail’s pace. This slow progress has not significantly accelerated even with major national and regional efforts in a number of nations to promote the use of the electronic health record (EHR), computer-based physician order entry (CPOE), electronic prescribing, and the personal health record (PHR), which are important substrates on which CDS can operate (and for which the prospect of CDS itself is a major driver). Some capabilities have made their way into commercial health information system products – examples include advice and warnings during CPOE, to ensure proper doses, avoid harmful interactions, or warn about allergies, the provision of alerts to providers when an abnormal laboratory result is found, and the use of order sets, or groupings of orders, for specific clinical problems and settings, such as coronary care unit admission or post-operative care after a hip replacement. Nonetheless, CDS usage remains spotty at best, most prevalent but by no means ubiquitous at academic medical centers, less so in community hospitals, and almost non-existent in office practice. Although the public frequently now turns to the internet for medical knowledge, automated decision support oriented to patients and consumers in terms of reminders, alerts, and patient-specific advice is similarly largely lacking.

What’s wrong with this picture? Is CDS perhaps not really a good idea? Are the requirements for wide dissemination and use beyond reach? Or are there initiatives that can be undertaken that can change the dynamic, and significantly boost adoption? This book is an effort to address this conundrum. Our goal is to examine CDS in detail from many perspectives – its history, the motivations for it, the technologies, the psychology and human factors concerning deployment and use of CDS, the sociology and organizational and management considerations that can lead to successful deployment, the financial and economic drivers and constraints, the marketplace and business opportunities, and lastly, the role of communal and top-down initiatives such as standardization and creation of infrastructure and sharable resources.

The premise is that if CDS is truly a good idea, then the sluggish progress in adoption and use to date can only mean that we are in need of a new approach. But to develop a new approach to a complex, multifaceted problem, one that would have a better chance of success than current incremental, uncoordinated efforts, the effort will require the participation of many stakeholders representing a range of perspectives. It is much easier in such situations to preserve the status quo, or to introduce minor tweaks than to take concerted action – since tweakers can tinker with an existing system without needing to build a huge consensus, but concerted action requires a major effort to promote a shared vision. Only by finding a way for these participants to come together with a commonality of purpose can this good idea be driven forward with the duality of courageous patience and impatience that it requires.

This book is thus an effort to develop a common ground for addressing this challenge. It should be of most interest to health care organization managers, policy makers and other senior leadership, payers, government funding agencies and foundations focused on health care delivery, medical informatics researchers and students, information technology development managers, information systems and knowledge product and service vendors, and clinical investigators and health care providers more generally who have interest in the issues of health care quality, safety, and cost-effectiveness.

Let’s look a bit more at the motivations for CDS and the challenge of aligning them. In this book we adopt a view of CDS as decision support aimed at individual patient-specific health care. It is advice and guidance offered by computers (more properly, information and communication technology) to aid the problem solving and decision making of health care providers, patients, and the public (i.e., including those not currently patients). CDS is in most views not only a good idea but an essential one. The most compelling reasons for CDS are to help practitioners avoid errors, optimize quality, and improve efficiency in health care. Many pressures fuel the need – the explosion of biomedical knowledge over the past several decades, the multiplicity of diagnostic and therapeutic choices available for patient care, the specialization and fragmentation of care, the time constraints on practitioners, regulatory and compliance demands, malpractice concerns, the increasing prevalence of multi-system diseases as the population ages, the spiralling costs of health care, the growing activism and involvement of individuals in their own health care, and the emerging capabilities for “personalized medicine” through genomics, biomarkers, and increasingly structured clinical phenotype data.

Although the descriptors used to characterize the above trends have a kind of desperate urgency to them which collectively suggest hyperbole, the fact is that they do reflect the reality of health care today. Because the trends and their consequences impact on different stakeholders in various ways, however, the combined extent of the need perhaps has not been appreciated to its full degree by individuals, as a result of which there has been little impetus for a broad-based effort to address it.

It is important to understand the differences in perspective of the various stakeholders, and to recognize their motivations and needs. For practitioners, despite the benefits offered, there are a number of reasons why CDS is not unequivocally endorsed. Providers often become quite expert in their own particular subject domains, keep up with the literature, and don’t feel a compelling need for computers to make recommendations. Despite this, they do generally accept the value of CDS to monitor their actions especially where the aim is to help avoid accidental errors; they welcome alerts for unexpected lab results – if false positives are kept to a minimum; they appreciate timely reminders for schedulable actions; and they take advantage of predefined order sets for frequent clinical situations they encounter. In general, and not surprisingly, the satisfaction with CDS by practitioners seems mostly related to the degree to which it is supportive, patient-specific and relevant, and provided in a way that doesn’t interfere with care or require inordinate additional effort and time.

In some circumstances, however, the use of CDS not only requires extra time and effort by the practitioner, but the benefits of its use aren’t seen as accruing to the practitioner or the patient. Examples are applications of CDS aimed at limiting orders for expensive tests and treatments. This purpose has been manifested in adoption of drug formularies, substitution of generic for brand name drugs, utilization review and utilization management, and requirements for preapproval/prior authorization from payers for imaging procedures, specialty referrals, or surgery as a function of clinical indication. These may be tolerated by practitioners as necessary evils but they are regarded primarily as interference with medical judgment and are hardly embraced. Patients also exhibit disdain for and distrust of such applications. So although a societal net benefit may be at play, it is difficult to align support of institutions, payers, providers, and patients in cost containment circumstances.

Difficulties such as the above have been true of many information system innovations in health care. Given the frequently tenuous acceptance or tolerance of such technologies by those who must interact with them directly, instances in which their implementation has been poorly done have been quick to be seized on by unhappy users and critics as examples of why the adoption of such systems should be resisted. Before an information system innovation is introduced, careful thought and experience with testbeds and pilot implementations are needed. Planning, both locally and at an organizational or enterprise level, needs to be directed at issues of how to motivate the participants, and how to make the innovation function effectively, including...