Many chief information officers will be physician executives.
Organized delivery system, capitation payments to providers, practice guidelines, and profiling of physicians' practice habits will all change the power structure in departments of information systems from one that favors financial processing for billing for procedures to clinical processing for clinical data collection, communication, and analysis to make health care services as efficient and effective as possible. If patients are to be identified electronically at all locations of care within an integrated delivery system, and if the findings of clinicians and treatment decisions are to be codified electronically for electronic storage and distribution by computer-based patient records, behind the clinical information systems must be enormous effort at standardization of clinical vocabularies and medical records. The growing importance of standardized electronic medical records will bring clinicians quite naturally to leadership positions in departments of information systems.
Many people are interested in comparing the effectiveness and efficiency of various ways of treating patients. As a society, we need to measure the relative merits of alternative ways of treating specific medical conditions. There are no direct measures of efficiency or effectiveness, because there are no measurement scales for efficiency or effectiveness. The measures we use, such as costs of care, charges for care, cure rates, mortality rates, five-year survival rates. patient satisfaction scores, and patient functional status scores, are all proxies for the concepts we want to measure - efficiency, effectiveness, or both. Increasingly important to clinicians and institutions will be comparisons of their effectiveness and efficiency to include, or exclude, them from alternative delivery systems such as HMOs, IPAs, PPOs, PHOs, and OWAs (other weird arrangements).
Profiling providers for inclusion in, or exclusion from, deliver systems already occurs. It affects subspecialists more than primary care physicians, who usually are more attractive to alternative delivery systems, and occurs more often in urban areas, where concentrations of all providers are highest. Until recently, profiling by insurers has been inclusive in nature, ruling out those few clinicians with unacceptable malpractice histories or inadequate credentials and those few facilities that do not meet the standards for accreditation of the Joint Commission on Accreditation of Healthcare Organizatioins. Most payers say they can't afford to be inclusive of almost all providers any longer.
In order to compare the outcomes of patients and measure the relative effectiveness and efficiency of specific treatment methods, we must control for each patient's prior probability of the outcome on which the comparisons depend. Before treatment, patients come to clinical trials with varying probabilities of the outcome of interest to researchers. Before they are treated, patients with cancer vary in their probabilities of surviving five years. Some of them suffer with cancers that are widely metastatic, while others are blessed with tumors still localized and eradicable with surgery and/or radiation therapy.
Therefore, comparing the outcomes of treatments for patients depends on fairly comparing patients' prior probabilities of the outcomes under study before treatments are introduced. This is as true for academic research designed to compare the outcomes of patients treated with alternative experimental drugs as it is when profiling physicians to compare the effectiveness and efficiency of their clinical practices. An endocrinologist who has filled his practice with svelte, young, athletic, diabetic patients whose blood sugars are controlled well with diet and exercise and who have no measurable end-organ injury from microvascular disease will measure far better outcomes for his patients, in terms of costs of care, rates of complications and hospitalization, and rates of referral to other consultants, than an endocrinologist whose patients are obese, elderly diabetics requiring complex regimens of insulin and already suffering with multiple end-organ injury. In order to compare fairly the performance of these two endocrinologists, we must control for the probabilities of outcomes among their patients before they began to treat them.
We can control for prior probabilities of the outcomes under study in one of two ways. Either we carefully control the patients entered into a clinical trial, to make certain that patients are comparable in all ways except for the treatments the outcomes of which are being compared, or we use statistical techniques to control for variations in patients' prior probabilities of the outcomes in question, using regression analysis performed by statistical software.
Health care organizations need health services research staff who know how to analyze clinical data for presentation to committees wrestling with clinical quality improvement initiatives, developing clinical practice guidelines, or profiling the practice habits of clinicians. Health services research staff need to work closely with information systems staff who create the data models, data dictionaries, and databases for data analysis. Far better for them to work side by side in the same department of information systems than to sequester them in separate cost centers where they will tend to work independently, and often competitively, duplicating efforts and producing inconsistent data resources and analyses.
Departments of information systems of health care organizations usually are not at the center of clinical data collection and analysis, in part because they are usually led by managers who are not familiar with clinical practice. They usually report to vice presidents of finance. Data collection and analysis are performed throughout health care organizations, usually by staff reporting to various middle managers in charge of marketing, human resources, finance, risk management, quality improvement, quality assurance, and clinical research. These fragmented services usually do not offer their organizations a standardized nomenclature for defining the findings and experiences of patients, or of risk-adjusting for the prior probabilities of patients' outcomes, or a standardized way of presenting data analysis to clinicians and managers. Each department head has his or her own analytical staff, and usually they do not establish common databases for analytical work.
Health care organizations now feel the pressure to invest in data warehouses for retrospective analysis, because clinicians' and managers' interest in data collection and ad hoc analysis is growing with increasing turbulence in the marketplace for health care services. There is increasing urgency to perform ad hoc analysis of health care data to respond promptly to requests for proposal from insurers and to demands for outcomes data from regulatory agencies and the media. In the absence of a corporate data warehouse containing all the financial and clinical data on patients, small, incompatible, inconsistent databases crop up under the guidance of data analysts in the various departments mentioned above, with little or no standardization of data definitions, data formatting, data modeling, query design, or risk adjustment among them.
Information systems have traditionally not played a leading role in decision support for health care organizations. Finance and accounting have played that role, with financial information systems giving them their data. That orientation must change to support clinical quality improvement, development of practice guidelines, and standardized profiling of clinicians' practice habits. No other department is better positioned than information systems to play this crucial role of maintaining and protecting the information resources of the organization, including transaction systems, communication networks, and data repositories for retrospective analysis of clinical and financial data. Departments of information systems must take on a sales orientation, and analyze the needs of their constituents for useful information and communication resources before their customers ask for them. They need to go after opportunities to help clinicians and managers improve the work that they do.
The key technological trend that is moving information systems to the forefront of clinicians' attention is the rapidly increasing power of personal and mid-range computers that, with graphical user interfaces and fast microprocessors attached to corporate networks,, can give clinicians windows to an electronic world of multimedia data about their patients - alphanumeric text, sounds, images, graphics and video - from any terminal in their health care facilities, in their offices, and in their homes. Now there is no technological impediment to collecting, storing, transmitting, and displaying every component of the medical record electronically on personal computers attached to wide-area networks of regional health care organizations. The personal computer workstation becomes the clinicians' power tool for access to patients' demographic and clinical data and to databases designed for retrospective inquiry into the treatments and outcomes of many patients. Databases for decision support are becoming the principal resources for clinical quality improvement and outcomes analyses for all health care organizations.
Departments of information systems already maintain the transaction systems for financial data processing. Usually they also maintain the information systems for clinical transaction processing, such as laboratory, pharmacy, and radiology systems. They manage the telecommunication networks that move data from servers to workstations and back again. But departments of information systems usually do not maintain the departmental data repositories, such as tumor, transplant, and trauma registries, used for recording the clinical treatments and outcomes of patients. Usually, departments of medical records are not part of departments of information systems. Medical records departments, when they move to electronic storage of medical records in computer-based document processing systems, usually do not include departments of information systems in their planning or implementation. They hire outside consultants to help them draft requests for proposal to send to vendors recommended by the consultants or identified in exhibits at national meetings of medical records professionals. Departments of finance usually do not involve information systems in procurements for cost accounting systems. They also tend to hire outside consultants to lead them through procurements of cost accounting software and only involve departments of information systems when interfaces are required between the newly selected cost accounting systems and the existing patient accounting systems.
All of these independent decisions made by autonomous operating departments lead to a polyglot of data definitions for key clinical concepts and later to difficulty in standardizing clinical nomenclatures to define the findings and treatments for patients. Not all clinical registries use ICD-9-CM codes for diagnoses and procedures. Some clinical departments, such as orthopedic surgery, need extensions of the ICD-9-CM methodology; some need nomenclatures to describe anatomic findings of patients, such as SNOMED III; some need unstandardized terms to describe clinical findings, such as measurements taken at cardiac catheterization; and some need data from their own idiosyncratic survey instruments that measure patient satisfaction and functional status.
Let me suggest a job description for a clinical chief information officer. The CIO, of course, must supervise and manage the organization that installs, implements, maintains, and upgrades electronic information and communication technologies. Usually, an administrative CIO does these things and not much more. A clinical CIO will also lead clinicians to successful data standardization, collection, and analysis for clinical quality improvement and outcomes management. The clinical CIO will create and supervise a division of clinical informatics devoted to developing expertise in data collection and risk adjustment to support profiling of physicians and research studies for clinical quality improvement exercises. The clinical CIO will teach clinicians and managers to understand the benefits, costs, and limitations of computer-based patient records and will lead the standardization of those records, including clinical departmental data repositories, such as tumor, transplant, and trauma registries, to support patient care. The clinical CIO will lead the development of teleconferencing capabilities to support telemedicine programs to more remote populations underserved by subspecialists. The clinical CIO will lead the training of administrative and clinical staff on personal computers to permit them to perform their own retrospective data analysis using risk-adjustment methods and databases standardized for the organization.
The key to successful electronic communication is standardization - for data, for software applications, for operating systems, for hardware, for networks. In fact, standardization of computer records is much more important than standardization of paper records, because humans can recognize patterns in text and on paper of various sizes and shapes much easier than computers can. Leading the standardization of clinical and financial systems to support computer-based patient records for on-line transaction processing and retrospective analysis is perhaps the most important and retrospective analysis is perhaps the most important role of the clinical chief information officer.
You might ask why health care organizations may not be better served with a layman who knows computer and communications technologies intimately and has supervised operations of information systems before as the chief information officer, and a clinician as clinical information officer, reporting to the layman and leading standardization of clinical nomenclatures, selection of clinical transaction systems, and data analysis for quality improvement and guideline development. My observation is that the tone and tempo of an organization, like an orchestra, is set by the conductor. I believe the conductors of organizational informatics of large health care organizations must know clinical work intimately. If this is the case, why are there not more clinicians as CIO? The rest of administration of a health care system may be anxious about a clinician CIO, because they have never dealt with one before. The CEO of a health care system may not want a clinical CIO, or any CIO, reporting to him because he does not feel competent to guide someone with training in technologies beyond his ken.
Those risk-adverse CEOs need to go into retirement, in my opinion, and clinical CIOs need to report to CEOs who are not afraid to learn what they need to learn to support and guide their CIOs to create information and communication systems that are becoming the most important assets, after the people delivering care, of the emerging health care systems they are paid to lead.
Health care organizations need clinicians in leadership roles, to lead rather than to advise administrators. The most important assets of health care organizations will be their information and communication technologies supporting electronic collection, storage, communication, retrieval and analysis of data on patients and operations. The clinician CIO can be a nurse, physical or respiratory therapist, pharmacist, or physician, but, while allied health professionals with training in information systems are invaluable for selection and implementation of clinical transaction systems, such as order entry and results reporting systems, physicians will more likely succeed in leading other physicians to adopt and use computer-based patient records, to standardize their data collection in the office to support computer-based patient records for inpatients and outpatient use, and to collect and analyze data from clinical processes and outcomes for profiling the efficiency and effectiveness of alternative clinical treatments. The table on page 39 lists some of the specific projects related to information and communication systems that many CIOs are wrestling with now, or will face in the near future, and my assessment for each project of whether a CIO with clinical training would have an advantage in managing it.
How have most organizations managed to cope without clinical chief information officers? They have not invested, yet, in clinical transaction and decision support systems that affect clinicians, especially physicians, in any substantive way. They have invested in telecommunication networks, financial transaction systems, and order entry, results-reporting systems that physicians rarely use. They have not actively invested in clinical decision support systems, such as expert systems to give physicians advice in selecting and dosing pharmaceutical agents for patients or systems to give physicians counsel on their differential diagnoses for challenging patients. They have not introduced telemedicine networks. They have not created, or acquired, data warehouses of clinical and financial data on patients for retrospective analysis of the quality and the outcomes of care. They have not introduced the use of risk-adjustment methodologies to control for the probabilities of patients. outcomes before treatment. Financial and marketing executives usually select case-mix systems without clinicians' active participation. They have not promoted the acquisition and use of computer-based patient records for physicians to use in their offices and in hospitals.
In other words, they have tended to avoid information and communication technologies that will directly affect physicians and have maintained traditional information systems departments concentrating on financially oriented accounting, inventory, scheduling, communication, and payroll systems. When physicians are directly affected by information or communication technology, lay vice presidents of information systems or chief information officers usually find clinicians to play the unofficial role of clinical information officer, to run interference for them and promote clinically oriented information technologies to clinicians. Shortly, almost all the investments made for information processing will be in clinically oriented systems, and organizations will see the wisdom o hiring clinicians with formal training in health care informatics as chief information officers. In a health care organization, where the purpose of information systems increasingly will be to support and expedite the efficient and effective delivery of clinical care to patients, it is inconceivable that chief information officers will not