Kenneth A. De Ville, Ph.D., J.D.
Testimonials to the capabilities and promise of handheld computers, or personal digital assistants (PDAs), in medicine are increasingly common in print, in clinics, in hospitals, and in medical schools. PDAs are designed to perform a wide range of clinical and administrative tasks for physicians and are taking the medical world by storm. In January of 2001, industry data indicated that 20% of U.S. physicians carried and used PDAs. That number is undoubtedly higher now. One report, titled “The Cure is in Hand,” predicted that handheld devices will “dramatically change the way that physicians practice medicine.” According to its authors, PDAs soon will be used for everything from scheduling to prescribing, from billing and reimbursement responsibilities to medical decision support . While the use of PDAs at the bedside is likely to bring much benefit to patients, society and the medical profession, the precipitous adoption and use of such a powerful technology may mask a broad range of professional and ethical issues.
When considering the adoption and use of a new medical technology, there are two potential classes of error: (1) clinicians and institutions may adopt an innovation too soon or wrongly; or (2) they may delay or fail to adopt a beneficial technology. As a general matter of professional and ethical propriety, a new medical technology should be employed if its use “improves the quality of care at an acceptable cost in time and money or if it maintains the existing standard of care at a reduced cost in time or money .” Benefits both to the patient and to society must be weighed. Patient and clinician preferences are relevant, though not necessarily controlling, when other factors (e.g. quality of care, economics, or overall social good) indicate that the new technology offers, on balance, clear advances over previous ways of practicing medicine and providing health care services.
In some cases, the new technology may offer only a minor benefit over previous and conventional practice patterns. Or, the benefits or risks of the technology may be uncertain or indefinite. In such instances, clinicians may choose to adopt, or to ignore, the innovation. In other cases, however, the overall benefits of a new technology impose a positive moral and professional obligation on the physician to weigh carefully and honestly the value of the technology and perhaps to integrate the innovation into his or her practice. Conversely, new medical technologies should not be used if the overall benefits to patient and society are not clearly demonstrated, or if the risk, time, expense, and resources expended on the new technology are not justified by patient gain. Finally, the use of some technologies may be justified only if they are used in a responsible and appropriate manner.
THE STATE OF THE ART: EVALUATION AND ANALYSIS
The use of PDAs may raise a number of important questions related to clinical and ethical propriety, the answers to which depend substantially on the type of functions for which these devices are used. For example, PDA software that transposes a physician’s paper calendar to electronic form using a handheld version of software merely substitutes for a physician’s hardcopy paper calendar. Confidentiality concerns aside, there is little to suggest that the benefits and risks of scheduling software installed on PDAs either mandate or preclude the use of this technology. Scheduling software may lead to some increase in both the clinician’s and, by extension, the health delivery system’s efficiency. If these economic benefits of such programs are significant, then it may be possible to argue that their implementation is ethically required in order to conserve resources for uses in other areas of the health care sector.
A second category of PDA software, data management programs, allows physicians to keep notes in the traditional history and physical (H & P) or Subjective, Objective, Assessment & Plan note (SOAP) format. In addition, such programs may allow date and time entries for current medications, labs, vital signs, and the results of a wide range of other diagnostic procedures. At first glance, PDA patient management programs appear to merely replicate the hard copy notations on patients that physicians traditionally have dictated, or taken on index cards or cocktail napkins. However, it is increasingly clear that such software can record more information in a more legible and organized form than any form of handwritten notes. This attribute has implications for the improvement of patient care and the efficiency of both the health care institution and the health delivery system at large. Handwritten patient notes, in the short term, sometimes can leave other members of the health team with gaps of information, which cannot be filled without time-consuming and difficult follow-up searches. More accurate and more efficient transmission of important patient information has the potential to improve patient care. The extent of these benefits remains uncertain. The promise of patient management systems brought to the bedside on PDA platforms is striking and sufficiently compelling to suggest that medical professionals and health care institutions may have an ethical and professional obligation to explore the prospects and impact of widespread adoption by clinicians.
MEDICAL REFERENCE PROGRAMS
Another category of PDA software, medical reference programs, allows clinicians to carry complete versions of classic medical literature, practice guidelines, and clinical drug databases. Medical reference software, like patient data and management programs, has the potential to directly improve patient care on a daily basis. Data management programs help the clinician recall and organize the enormous amount of patient information generated by modern diagnostic technologies. Medical reference programs, in PDA format, allow bedside or examining room access to the vast amount of general medical knowledge generated in the second half of the twentieth century. The human mind can only hold and recall a limited percentage of the scientific, diagnostic, prognostic and treatment information now available. While this information is theoretically accessible, the practical difficulty in consulting these sources on a patient-by-patient basis has limited its widespread use. The same information, in PDA format, can be used more frequently and, to greater advantage, than ever before.
As long as traditional medical reference tools are reproduced in total electronically, it is difficult to oppose their widespread use. However, there is still room for some caution. Clinicians and institutions should be wary if the electronic medical reference tool contains less information than the previous hardcopy version or if it is organized in a substantially different manner that may subvert easy or familiar access to information. In these instances, it is incumbent on the user to be aware of the gaps in the electronic version and be certain that he or she has a working knowledge of the software before employing it in practice.
Some observers may worry that ubiquitous electronic availability of medical reference sources may lead to a degradation of physician standards and skills, and that clinicians will grow to rely too heavily on the PDA reference programs. Perhaps. But how great a danger do these fears represent if such technology becomes ubiquitous and clinicians always have access to their handheld computers? Medical reference texts in electronic format do not directly threaten to supplant physician judgment and analysis. They merely provide information that is already available in a more convenient form. If physicians begin to treat electronic texts as the final and fixed word on clinical matters, then these programs may pose a risk of danger; but if clinicians give electronic versions of standard research tools no more or less respect than they have given their previously available hardcopy versions, there is little reason for concern. What little danger does exist is almost certainly mitigated by the benefit of having immediate access to hundreds of pages of standard literature that should help both strong and weak physicians practice better medicine.
Several available pharmaceutical reference programs begin to blur the line between pure reference sources and decision- assist programs. These pharmaceutical reference guides have the potential to improve efficiency and protect the safety of patients by preventing prescription error. More specialized software promises other benefits. One handheld application helps clinicians choose the most appropriate antibiotic therapy, an increasingly difficult task given the range of potential available remedies. Advocates contend that such programs may help clinicians decrease their reliance on broad-spectrum antibiotics and the looming threat of widespread anti-microbial resistance. Many of the pharmaceutical software programs provide for regular, even daily, updates on available medications, as well as indications and adverse events.
The availability of electronic pharmaceutical reference tools and their ability to improve care and limit errors may make their careful integration into everyday clinical practice an ethical and professional obligation. Again though, caution is in order. No such program provides an inclusive listing of all medication and all relevant information. All do not provide updates on late-breaking information at the same rate or easy access. Thus, while there may be a mandate to investigate and take seriously the potential benefits of adopting this technology, there is a corresponding obligation to understand the limitations of the various options and choose the most appropriate software. Finally, if such programs are employed in clinical practice, clinicians will have an ongoing responsibility to update information, when it is available, and to supplement the software in the areas in which its guidance is limited.
DECISION SUPPORT PROGRAMS
A final category of medical software available in PDA format is decision support programs. These programs provide both more and less guidance than is available from a mere digitalized reference book: more in that they frequently include diagnosis, treatment and medication algorithms, decision trees, outcomes data, evidence-based guidelines, and predictive calculations; less in that they do not include all the information contained in standard medical reference texts. Some such software allows clinicians to enter patient specific data and test results and generates both the reliability of the particular diagnostic test and the overall probability that the patient is suffering from a particular malady. Newsweek magazine proclaims that one such program, Inforetriever, “doesn’t just enhance their [physicians’] efficiency . . . it can improve their decisions .”
Despite this enthusiasm, many commentators have argued convincingly that we are unlikely to develop a computerized system that can even approach the diagnostic ability of the human physician, and are therefore encumbered by the greatest number of uncertainties and potential hazards . Diagnosis is often more than a mechanical, algorithmic function. Instead, as Randolph Miller has explained, medical diagnosis: “ . . . represents a sequence of interdependent, often highly individualized processes: elicitation of initial patient data; integration of the data into plausible scenarios regarding known disease processes .” For a number of reasons, computer programs are unlikely to fully replicate this sophisticated and in some ways still mysterious diagnostic exercise. Diagnostic programs, however, need not be infallible to provide some benefit to patient and society. They may serve as a guide to a clinician’s thinking through a problem or a memory spur to a physician who has momentarily forgotten the details of a particular diagnosis or set of symptoms.
Instead of diagnostic infallibility, the more appropriate test of whether a particular decision-assist technology should be employed is whether it provides patient and societal benefit that exceeds the risks and burdens that it carries with it. For a clinician to know when it is advisable to use, and in some cases rely on, diagnostic software, he or she would have to be sensitive to the limitations of the technology so that its guidance can be ignored. At this stage of development, most rank-and-file
physicians will not be capable of this insight and, for them, the decision-assist program loaded on their PDA will remain very much a “black-box.” The clinician user would have to have a grasp of the various databases on which the digital-consultation bases its conclusions. If the program contains information that is dated, incomplete, or limited in some other respect, the advice generated may be inapplicable to the case at hand but the physician would not know it. At a deeper level, a physician using a decision-assist product installed on a PDA has a duty to understand the nature of the algorithm used in developing the program and “the ways in which it might fail, both due to inherent theoretical limitations and due to the flaws that might occur during the process of implementation .”
An insightful and experienced clinician would presumably be able to identify those times when a decision-assist program provided incorrect or suspect recommendations and thus reject them. Not all clinicians, however, are insightful and experienced. In fact, it is possible that the physicians most likely to take advantage of the new wave of computer technology will be medical students and inexperienced physicians who have yet to develop fully their analytical, problem solving, and clinical skills and thus may be less able to recognize blind spots in the various available software.
Given the difficulty of fully understanding the implications of both the data and analytical shortcomings of these programs, a comprehensive training program seems a reasonable protection against the potential over-reliance on an inherently limited technology, which is proliferating without oversight. Other measures might include ongoing assessments and clinical trials of the safety and efficacy of decision-assist programs when used by physicians of differing skills, experience, and specialties.
The foregoing discussion should be considered a guide for physician and institutional self-evaluation—a framework to identify relevant factors to be considered when making adoption decisions on a case-by- case basis. Hopefully, such critical insights will temper inflated expectations regarding the clinical use of handheld computers, slow or stop problematic applications of the technology, and identify those uses for which resistance to change is no longer professionally and ethically justified. [This article is a shortened and revised version of De Ville, K. “The Ethical and Legal Implications of Handheld Medical Computers,” which appeared in the Journal of Legal Medicine 2001; 22:447-466.]
1. Fisher, J. and Wang, R. “The Cure is in Hand.” http://www.wrhambrecht.com/research/coverage/ehealth/ir/ir20001019.pdf (last visited June 5, 2001) (market report from WR Hambrecht + Co.).
2. Miller, R.A.;Shaffner, K.F.; and Meisel, A. “Ethical and Legal Issues Related to the Use of Computer Programs in Clinical Medicine.” Annals of Internal Medicine 1985; 102: 529-37.
3. Cowley, G. and Underwood, A.“ Finding the Right RX.” Newsweek 1999; 134: 66-75.
4. Yu, V. L.; “Conceptual Obstacles in Computerized Medical Diagnosis.” Journal of Medicine and Philosophy 1983; 8: 67-75.
5. Miller, R.A.; “Why the Standard View is Standard: People, Not Machines, Understand Patients’ Problems.” Journal of Medicine and Philosophy 1990; 15: 581-91.