During November 12-15, 1996, I was a delegate to an Expert Consultation Meeting, “Telecommunications in Health and Healthcare for Latin America and the Caribbean ,” held in Washington, DC by PAHO (the Pan American Health Organization), a part of WHO (the World Health Organization). The meeting was organized by the Health Services Information Systems Program of PAHO’s Division of Health Systems and Services Development.
I was assigned to Workgroup 3B: “Needs Assessment.” The other delegates in that group represented Chile, Mexico, the USA, Argentina, Uruguay, and Nicaragua. After talking with them for the first 3 days of the meeting, I wrote up a draft of my personal ideas for possible inclusion in our report back to PAHO. The body of this post contains my draft report. I must emphasize that these are my own views and do not necessarily represent official positions by PAHO or any of its member states, nor are they necessarily endorsed by the other members of Workgroup 3B. (PAHO’s logo artwork above, is used with their permission, recognizing my efforts at the meeting.)
I do thank everybody for what they taught me at that meeting.
(Subsequently, two of the delegates — Sandra Mejía Mendoza, from Nicaragua, and Alvaro Gaynicotche, from Uruguay, worked with me to develop a paper, Healthcare Systems Process Reengineering for Developing Countries: A Report to IMIA Working Group 9, that I presented at MEDINFO 98 in Seoul, Korea.)
The structure of the health telecommunications in a country should follow from the structure of the country’s programs for the care of its citizens, and therefore we have to look first at health care delivery itself.
Different countries have diverse models of care, derived in part from their national culture and aspirations, in part from their resources, and also in part from accidental historical factors. It is not the role of PAHO to develop and implement solutions, however brilliant in the abstract, for a country from the outside. Rather, PAHO’s mission is to find ways to adapt general tools to local circumstances so as to use existing political energy and existing local resources (which in many cases are as good or better than can be imported) to help enable countries to provide for their own welfare as they see fit. The key word is “enable.”
Despite the diversity of national goals, we can perhaps single out a way of stating a feature of health care systems that is common to all, but articulated clearly by few if any. Historically, in most countries health care has been organized around the functional divisions of medical specialties; around the professional distinction among doctors, nurses, and other health providers; around distinctions between health care, community health, and social and environmental health factors; and its organization has tended to assume that health is the job of professionals, with the consumer as a passive recipient.
Health care organized on these lines has done well, in many ways miraculously well, and we need to be grateful for it. Still there are ways that it has done less than it could. In particular, there are substantial inequities in the care available to poor, to rural, and to inner city, individuals. Further, the cost of health care as a whole, under the traditional organization, has been shown to be impossibly expensive, even for nations considered rich.
We need to articulate a model of health care (and of the telecommunications structure that could support it) that would help countries in the search to avoid wasting their existing resources (in personnel, in facilities, in communications infrastructure, and in information) and instead to use these resources effectively to provide equal access to wellness and to quality health care for all their citizens.
We propose, or endorse, a new model with the same goals as the traditional one, but (we feel) more likely to achieve them. The basic model that we suggest is that of structuring care around the longitudinal process of the patient rather than around the functional divisions of the care provision system. The object of the health care system should be to follow individuals so as to ensure that
- they remain healthy to begin with and do not require care;
- if they do develop acute illness, they are cared for well, rapidly and cost-effectively;
- if they are chronically ill, they are made comfortable and the progress of their disease is slowed.
To be effective, the central concept of the new model must be the need for a continuous, well-coordinated, connection among the services provided to each individual. An individual, perhaps poor, perhaps rural, perhaps inner city, has a primary care provider, perhaps a physician but perhaps a nurse or other health worker. We must ensure that the primary provider has the tools to follow, coordinate and ensure the quality of the all the processes in which care is given the that individual.
The functional divisions of the traditional system are justified by the impossibly large amount of knowledge needed in subspecialties, by economies of scale in the use of specialized equipment, and by considerations of triage. Nonetheless, these distinctions have been carried too far and have become pernicious: in poorly coordinated, redundant and ineffective cooperation among providers for the same patient, and also in the artificial and unhealthy assumption that the health care system is responsible for a person only after he has become sick. Sickness is a waste in the quality of the individual’s life, in his potential contribution to society, and in the social resources needed to make him well, or partially well, again. We need to avoid sickness in advance.
It is an additional corollary of this concept that the individual (and the community) has to be involved as an active participant in maintaining wellness, and as an active partner with the providers that are caring for him.
We do not propose scrapping the traditional system, but rather viewing it differently and also embedding it in a social program with a larger vision. The details of how this is done are to depend, as we said above, on each country’s needs, culture, and self concept, and on their methods, socialized or free market or mixed systems, of reimbursing providers and of ensuring quality of care.
To extend present health care systems to achieve this goal of individual and community wellness will not be easy; it will pose in each country a systems delivery problem of considerable difficulty. It is the role of health telecommunications, now called telehealth, to make the solution to these systems problems feasible. What we really are requiring is far greater cooperation among providers, far greater and easier access by each provider to the tools of his profession, far greater ability by individuals to be informed participants in decisions affecting their health and to comply with care plans, and a far higher quality of data and tools available to community groups and government agencies seeking to plan health delivery. These are the jobs that telehealth can do.
Certain aspects of telehealth, such as high bandwidth real-time video consultations, have a high appeal to the imagination; and these tools are indeed very valuable, when called for and when financially and technically feasible. But they are only some among the many components of telehealth. The main action is in relatively less expensive low and medium bandwidth applications. A country will adopt that portion of the telehealth toolset that fit its needs and capabilities. The role, and the economic justification, of telehealth is that it is a fundamental enabler in a nation’s effective use if its resources to bring a high quality, longitudinal, individual-oriented, program of wellness to all of its citizens.
In the following paragraphs we have tried to articulate the need for various components of telehealth, to describe them in useful but not too great detail, and to point out what some of the requirements would be to deliver them to their appropriate users.
General Needs and Considerations
Clinical Information Systems
Ultimately a clinical information system would have to include a complete Computerized Patient Record. In the final stages it would have to be in multimedia format, preserving the integrity of the kind of presentation needed for each of its components. It is, however, by no means necessary that it reach this level immediately: there are many very useful things that can be done with simple, low bandwidth, data.
For the record to be usable for the purposes described in our preface, completeness and fanciness are less important than that its distribution be in nearly real time and include all relevant clinical decision points, so that primary care physicians can effectively supervise and collaborate in the quality of an individual patient’s care as the patient passes through possibly multiple stages of consultation or through the stages of hospital care. As the patient is handed off, each new specialist needs to be able to see the previous stream of comments, decisions, reasons for decisions, and care plans, and to be able in turn to collaborate with the primary care physician. By “nearly real time,” we might mean by actual teleconferencing over a high bandwidth telelink or high bandwidth Internet connection; or perhaps over a slower, medium bandwidth World Wide Web connection. However, it might be equally effective and more efficient in the use of the providers’ time to have a much slower data and comments whiteboard over a text only telephone link: this way the providers can interact asynchronously, according to the requirements of their other duties.
Another important consideration is standardization of data. There are very many steps, electronic and terminological, of standardization required before the Computerized Patient Record can be shared among divers providers and also serve the research and systems delivery needs discussed below. Many of these problems can be obviated by planning standardization in advance. There may be a role for multinational cooperation, coordinated perhaps by PAHO. However, experience has shown that complete, theoretically ideal, standardization is a task that committees may spend years on and never complete. Therefore, it is important that useful projects not be held up. It is also clear that different kinds of health delivery operations do have legitimate conceptual differences, according with their very nature, and therefore a better approach sometimes is to allow these differences and to plan for means of translation and inter-operation.
We cannot overemphasize that it is not nearly good enough to have a computerized version of the conventional clinical record: the features that are essential for its role in the new view of care are
- nearly real time interchange among cooperating providers;
- comprehensive information about decision points: what information was available when the decision was made, which decision was adopted, and what was the outcome;
Clinical Decision Support Systems
The health providers in any country are a fundamental national resource. A country needs to nurture its providers. We need to support providers to do a better job in the face of their increasing workloads, the multiplicity of services that they sometimes need to provide for the same patient at once, and the impossible acceleration of the state of knowledge worldwide.
One method is by allowing the provider, or the patient, to consult with specialists via telemedicine, as discussed below. For economic reasons, however, this will inevitably be less frequent than using information technology directly to support the provider.
A great variety of such techniques is available, and dogmatism in the choice among them is simply inappropriate: national styles are too divergent, the needs of different types of patients (for example acute versus chronic) are too different, and the need for improvement by innovation too great. We discuss here a selection of these types of support.
While there are some disease conditions that reflect a simple model in which diagnosis, treatment and recovery fall into neatly separate phases and in which there is a unique answer in each phase, many conditions are not so simple. Patients may present with multiple symptoms and conditions, and with histories that strongly affect thinking about present diagnosis and treatment. Providers may start with a differential diagnosis from which they progressively rule out possibilities, perhaps only arriving at a final answer, or final answers, after treatment is completed. In parallel with this process, providers need to treat symptomatically, optimizing the patient’s condition in the light of standard objectives and known physiology.
When we also take into account the new requirement of basic, ongoing, health for the patient before he becomes ill, then we see even more strongly that a process view of medicine has to exist alongside the traditional question and answer paradigm.
It is also fundamental that we recognize a dilemma. Standardizing practices can help to ensure that the best treatments are widely disseminated and to remove treatment inequalities. However, at the same time, the realities of statistical variability prevent us from uniformly determining the best treatment for all conditions, and, also, basic and clinical research are constantly improving the interacting components of the state of the art. At the same time that we push for the benefits of standardization, we need to take active measures to ensure that we derive the benefits of decentralized innovation by individuals. Society is not always right, even for today; and we need to move on into the future.
Therefore, we need in part to disseminate treatment guidelines and standardized practices as determined by competent medical organizations. However, we also need in parallel to disseminate permissive technologies that enable initiative by practitioners attempting to cope with ongoing processes in the concrete context of individual patients’ lives. For example, we might attempt to guide providers in testing and the formation of differential diagnoses, rather than guiding treatment itself. Other suggestions are among the following.
Providers need comprehensive searching capabilities of the medical literature. It is imperative that intelligent indexing schemes be developed that will quickly lead providers past the theoretical and basic science literature and to the applications that are relevant in the context of the particular patient.
Guidelines and Clinical Pathways.
As we mentioned above, there are some disease conditions in which medical authorities could prescribe standard treatment guidelines based on scientific knowledge. In these cases, informatics provides a way to disseminate these recommendations nationwide, and also, by means of intelligent reminders in the computerized patient record system, it can ensure that providers, harried by multiple duties, always remember these guidelines.
Evidence based medicine.
Even in cases where there is not enough scientific agreement for medical authorities to agree on a unique guideline, we can improve quality of care by bringing summaries and references of the current state of scientific debate to the attention of providers seeking to help their patients.
Drug reference and Treatment interactions.
It is a waste to have providers using therapy that is less than the best, simply because of delays in disseminating information and difficulties in using it. These fundamental tools can be constantly updated and, using intelligent search techniques, made available electronically.
Physiologic models and calculations.
The quality of clinical decisions can be improved greatly if providers, instead of relying only on the raw data furnished by testing the patient, work the raw data up according the textbook equations of physiology. In practice these workups often require the provider to have a good memory for elaborate formulas and have the leisure for much hand calculation. However, it is easy to program such workups even into simple hand held computers.
There are many diseases in which expert medical opinion is deeply and bitterly divided over the merits of treatment options. Also it is often difficult for professional bodies to define treatment protocols that precisely fit the vagaries and individual exceptions found in practice. In contrast, it is much easier for medical authorities to define the circumstances in which a certain workup should be done.
It may be difficult to tell the provider, “We want you to treat in such and such a way,” but easier and much more effective to tell the provider, “We want you to consider such and such factors in making your decision.”
Formal methods of decision support.
There exists a wide variety of methods in the literature, and it is not the business of this document to pass judgment on them. However, we do note the range of such methods available and recommend that designers of systems consider their possible adoption, and also consider that individual users of the system may wish to experiment with these methods and that designs should allow the freedom to do so.
There is a formal method of decision theory formulated in terms of influence diagrams and probabilities. There are artificial intelligence techniques including rule based systems, fuzzy logic, and neural networks. There are statistical methods using bayesian analysis.
A familiar way in which providers reason is by comparing the present case with previous cases. This style of reasoning, which can be formalized in terms of the artificial intelligence technique called Case Based Reasoning, is in principle excellent. However, informal uses of it are not always as good as they could be. Use of this method is hindered by limitations in the provider’s experience, which may not be broad or deep enough. It is also limited by the provider’s memory. It is colored by subjectivity and by the statistical variability within that provider’s experience: one spectacularly bad experience with a treatment can outweigh the memory of several quietly successful ones.
A basic way that a nation can support its providers is by making available a collective library of cases, supported by artificial intelligence tools. The library can be built upon the Computerized Patient Record, provided that the record accurately reflects decision points and outcomes. This case library can help in several ways: by warning about possible side effects or pitfalls, by ensuring that therapeutic objectives are deep enough, and by reminding the provider about the full range of therapeutic options available. It is an enabling technology in that it fits well with the reasoning that providers often use.
The ability to bring remote consultation by specialists is a fundamental means of bringing equal access to care to all citizens.
Continuing Medical Education.
Providers outside of the high visibility centers are often dispirited about the deterioration of their skills and knowledge over time, as the state of medicine advances. Teleconferencing and interactive electronic books are an important method for helping these people to improve themselves and to do a better job for their patients. It is important that a country find better use for the resources that it already has.
Research in Treatment Modalities.
Nations must learn to view data as a national resource. There is no reason to continue using therapies when data exist showing they work less well than alternative possibilities. Therefore, having efficient means of using existing data to decide these questions is a way in which a country can use its resources to help its people.
The essential characteristic that can make data useful to countries seeking to concentrate their efforts on the better treatments is the one mentioned above, namely keeping accurate records that describe for each patient the decision points and the outcome.
Controlled clinical trials.
Controlled clinical trials are the gold standard of the scientific method for determining which treatments are the best. The paper based technology whereby this is currently done is tremendously expensive. The cost of this method could be considerably reduced (although still high) and thereby the method could be more widely useful, if electronic means were used to make the mechanical parts (randomization, data collection, data monitoring) a readily available part of the Computerized patient record. This would save the high costs associated with separate parallel activities of abstracting records to paper case report forms, auditing the forms, entering them in an analytic data base, and then edit checking.
Adaptive randomization to treatment.
While controlled clinical trials are the gold standard, their high expense and the need to recruit a large enough sample size mean that their use is very restricted. Even in diseases where we cannot recruit enough patients to reach a standard of scientific validity, we have an obligation to use the accumulating data in our possession to assign patients to the treatment most likely to help them. Statistical theories of adaptive randomization have been developed that would allow this to be done. Their implementation would require real time access, perhaps by telelink, to a computer program at time of randomization.
Personal Health Information Systems.
The idea that the national model of health care be centered around the needs of individuals requires that individuals be empowered with tools that enable them to participate as active partners in managing their health, their healthcare and their environment.
Population Data Structures.
Government bodies need aggregate data to support systems planning nationwide. Essential tools for this function are Geographic Information Systems (CAS), Complex Adaptive System (CAS) modeling tools, Decision Support and Influence Diagrams, and a variety of mathematical, statistical and economic tools.
Similar tools can enable governments to assess the impact of economic, social, and environmental conditions on the health of its communities. Satellite data may be an important additional means of gathering relevant data.
Connected, longitudinal care centered around individuals implies systems delivery by the public or by the private sector, or by a mixed system, as suits the individual country. Responsible bodies need access to the same sorts of population data tools as government regulators and planners. They also need software tools specifically for systems design. A notable example would be Business Process Reengineering (BPR) tools.
We need to move health interventions upstream by systematically enabling individuals and communities to control preventatively the cognitive, social and environmental preconditions of health and enabling them to exert demand management of health care as a counterpoise and complement to supply management by providers.
To support this change we need a comprehensive set of tools responsive to and usable by individuals, providing data, knowledge and analytical capability enabling them to have better input to health care decisions for themselves and for their communities.
A “community” can be a geographical and political unit — a town, a city, a county, or a state —, but can also be organized along other lines: a group of people served by the same Hospital or clinic, a group of people with a common disease, a group of people affected by a common environmental problem, a group of people in the same socioeconomic group.
The software should help “community groups” at several levels: official government or institutional provider policy-making organizations, unofficial grass-roots organizations, or simply small groups of individuals seeking to make an informed input into policy decisions.
Administrative Information Systems.
Government bodies need aggregate population data and data on individuals to supervise and regulate providers under existing systems. The providers, particularly institutional providers, need access to the same data in order to assess their own performance. Consumers and community groups, and also employers if they purchase care for their employees, also need this data. All groups will need appropriate assessment software.