benefits

INFORMATION POLICY FOR THE U.S. HEALTH SECTOR:
ENGINEERING, POLITICAL ECONOMY, AND ETHICS

SYSTEMIC BENEFITS

'Administrative simplification'

Individual business tasks, such as accounting or resource control, have long been undertaken by computers in hospitals and large clinics. Such systems were typically unable to communicate with those of other departments, however, much less computers at other institutions. Isolation necessitated costly multiple entry of common data, redundant storage (with inevitable inconsistencies), periodic conversions to and from paper-based records, and highly variable capacities for information retrieval in times of need (Institute of Medicine, 1991; General Accounting Office, 1991). Resource tracking, billing, and reimbursement are better integrated within institutions now, and more effectively computerized. So are administrative functions generally. Yet inter-institutional data flows are hindered by a lack of standardization (General Accounting Office, 1993; Workgroup for Electronic Data Interchange, 1992).

The US relies on a large number of public and private entities for financing of health services, each with its own complex requirements for reimbursement. While mergers and consolidations under managed care have reduced the number of interacting entities, it has exacerbated matters with its own layers of bureaucracy (e.g., such as pre-treatment authorizations). Such a structure is inherently costly to administer – the share of US expenditures devoted to administration is variously estimated at one-fourth to one-fifth of the health dollar. It is hoped that a continued movement from manual to electronic claims processing will reduce these costs. Insurance company experience with electronic transactions suggests substantial efficiencies related to billing, reimbursement, claims tracking, reconciliation and similar business functions (Institute of Medicine, 1994; Workgroup for Electronic Data Interchange, 1992).

Automated administrative records may also improve monitoring and detection of fraud, for example by flagging unusual utilization patterns or cross-checking treatment records with payment data (Office of Technology Assessment, 1995a). However electronic regimes also present new opportunities for malfeasance, potentially on a very large scale, and some commentators believe the net effect will be adverse (see e.g., Sparrow, 1996; Raab, 1996), absent a strong regime of administrative surveillance. Aside from the additional privacy risks, the costs of such policing will reduce the net savings from electronic conversions.

Clinical Care (patient-centered services) Paper remains the dominant storage medium for patient clinical data in most settings. Is represents a substantial volume of record-keeping. The General Accounting Office has estimated that some 34 million annual hospital admissions and 1.2 billion outpatient visits generated the equivalent of 10 billion pages of clinical records (General Accounting Office, 1993). Administrative and evaluative/research information is still typically recorded separately, or "mined" (extracted manually) post hoc from such paper documents. Computer-based patient record (CPR) systems, while focused on clinical care, are designed to generate information to support all the functions of the institution through linkable, shareable data. This "collect-only-once" feature of CPRs offers great potential savings over present redundancies (Institute of Medicine, 1991).

Patients receive several near-term clinical benefits from CPRs. First, linked electronic records are potentially accessible immediately, whatever the time, location or situation in which a person presents for care. Moreover, such records can potentially contain a complete health history, regardless of where or when care was previously rendered. This is a distinct advantage for a mobile population, which customarily receives health services from many different sources. Second, electronic medical records can be displayed in forms that more clearly highlight important details, compared to one or more paper files of uncertain organization. Both benefits have obvious utility in emergency situations. Eventually, computer data management and decision-support tools may routinely examine information in electronic medical histories, supplementing the evaluations of human care-givers in real time to provide improved clinical decision-making (Institute of Medicine, 1991; American Medical Informatics Association, 1995). Prototype electronic medical record systems in hospitals, focused on such functions as physician order-entry, have shown clear benefits, in both cost saving and improved decision-making (see e.g., Pestotnik et al, 1996; Evans et al, 1996).

Research (outcomes and effectiveness) Textbook models of medical technology development envision a linear, orderly progression from innovation to evaluation to diffusion. Randomized controlled trials are the paradigmatic tool to establish safety, efficacy and cost-effectiveness – after which regulatory approval is gained and routine clinical use begins. In reality, evaluation and diffusion commonly overlap and feed back on each other through the service lifetime of a drug, device or procedure (Institute of Medicine, 1990). Practitioners find new indications for use after release, for which no prior testing is performed. Development and refinement occur in day-to-day practice. Rare side effects and contraindications appear in larger populations that go undetected in small scale trials. Long-term effects emerge, but only with the passing of many years. Future research databases, extracted and aggregated from individual clinical/administrative data, would aim to provide an effectively universal regime of on-going evaluation at a comparatively modest incremental cost (Institute of Medicine, 1991, 1994).

Variously called "outcomes" or "effectiveness" research, such efforts would embrace broad studies of provider practice patterns and the distribution of health services, in addition to focusing on particular interventions. Unlike "primary" data collection efforts, where control groups and randomization can be used contemporaneously, such "secondary" analyses of data collected for other purposes must rely on statistical methods to eliminate bias. Techniques for doing so are well established, and much of the best current health services research relies on retrospective use of secondary data (Institute of Medicine, 1994). Erroneous/biased data remains a cause for concern, particularly in a world of fully-integrated electronic records. Such systems commonly include one-time-only data collection designs, from which all subsequent clinical, administrative and research translations flow. While efficient, this may increase the chances that biases from one application – such as "upcoding" of diagnoses to maximize clinical services reimbursement – will percolate through to other uses.

Public health (population-centered services) Public health, as its name implies, traditionally focuses on larger patterns of disease and disability in populations, in contrast to the patient-centered orientation of clinical practice. While it is common to view this century’s decreases in mortality and morbidity as achieved by the health system’s advanced technologies, much of the gain has come from rather pedestrian interventions of population-based public health, such as cleaner food and water sources (Folland, Goodman and Stano, 1993). Public health "population surveillance" is undertaken via a broad array of local and state public health agencies, clinics, laboratories and university-based schools of public health, collecting data from a wide variety of sources (Institute of Medicine, 1988). While such efforts include a focus on episodic threats of particular concern – e.g., AIDS, TB – the broader aim is on-going monitoring of all threats to population health – e.g., smoking, substance abuse, pollution-related conditions. Such epidemiological monitoring allows better planning of on-going health resource needs, as well as the early identification of new, previously unknown threats.

As for clinical research, aggregations of data from EMRs could provide an effectively universal regime of population surveillance at a relatively modest cost. Present monitoring capacities, based on a hodge-podge of reporting laws for various diseases and hit-and-miss screening for others, are widely judged to be inadequate to screen the great variety of health risks facing modern, mobile populations (Institute of Medicine, 1988; Gostin, 1994).