Research and Evaluation in Primary Care
Primary care can be investigated, and almost certainly improved, using scientific methods. The committee reached this view through testimony at public hearings, its workshop on the science base of primary care (Donaldson and Vanselow, 1996; Green, 1996; Povar, 1996) and a review of published literature. Although primary care research overlaps the field of health services research (IOM, 1995), it has certain special facets and concerns. As conceptualized by this committee, therefore, the primary care research enterprise was thought to be in a fledgling state—long on potential and short on actual accomplishment.
Funding and infrastructure to support primary care research stand in sharp contrast to the organized commitment to advancing knowledge in various subspecialty areas of medicine, typically using the methods of biomedical research and clinical investigation. This has three important ramifications. For one, current clinical research may have little to offer to primary care clinicians,1 as evidenced by the observation that "[a]lthough primary care practitioners can use some of the knowledge generated by [specialty-oriented] research, in fact, most of it is not relevant to primary care because of its focus on singly developed diseases, carefully selected patients, and the reporting of strictly physiological outcomes" (AHCPR Task Force, 1993, p. iii). For another, lessons from well-done primary care research are not available to inform the larger picture of health
By primary care clinician, the committee explicitly means physicians, nurse practitioners, and physician assistants, as discussed in Chapter 2 and the committee's interim report (IOM, 1994a). More generally, in referring to primary care, the committee means to imply primary care as defined in Chapter 2.
care organization and delivery. Finally, this paucity of primary care research and development leaves primary care insufficiently prepared to confront the challenges and opportunities inherent in the committee's definition.
One broad explanation for the mismatch between the bulk of clinical care and the bulk of health-related research is the misperception that primary care is already sufficiently understood (Nutting, 1996). Discussions in the workshop and elsewhere during this study made clear, however, that primary care is a distinct and quite complex field, that it is inadequately described to and poorly understood by the broader health care community and the public generally, and that it is apparently an important source of variation leading to different health (or disease) states and clinical management strategies. For all these reasons, the committee determined that setting out a coherent program for research and evaluation in primary care would be an important contribution of its report.
In this regard, the committee took note of Starfield's position (Starfield, 1996) that primary care research is "research done in a primary care context." Starfield argues that it is a fallacy to conclude either that primary care research can be done in anything other than primary care settings or that information purportedly about primary care that is drawn from research not done in primary care settings is a priori valid for the primary care clinician. With those cautions in mind, one can draw a reasonable inference that the nation has engaged in little primary care research.
To help redress this imbalance, this chapter explores the need for a primary care research infrastructure and identifies key areas of primary care research that warrant high-priority attention. In the first main section, the committee discusses four topics relating to the necessary infrastructure for primary care research and offers four recommendations designed to overcome existing barriers to such work and foster a stronger framework within which a broad range of studies might be conducted. In the second main section, the committee identifies high-priority areas of research. Although these topics are often in the health services research arena, they are highlighted here for two reasons: (1) to draw attention to the core elements of the committee's definition of primary care, and (2) to underscore the importance of conducting much of this work in settings that deliver primary care as conceived by this committee. The chapter ends with some commentary about the long-term impact of primary care research on the quality and costs of health care in this nation.
The committee's views on primary care research and an appropriate infrastructure in which to pursue it do not imply that the committee believes this part of the research enterprise should be separated from the rest of the research effort in this nation. Primary care research cannot be done in a vacuum. Rather, this work should be done in the context of developing data and insights for the entire approach to health care in this country—the realm of health services research (IOM, 1995)—so that a responsive, cost-effective, and high-quality system can be built in the ensuing years.
Support For The Infrastructure For Primary Care Research
The committee found challenging declarations of research agendas for primary care (e.g., Williams and Brook, 1978; Mayfield and Grady, 1990; AHCPR Task Force, 1993; Starfield, 1996); there is no lack of questions to be asked and answered. What does seem to be missing is a widely held commitment to the exploration and explication of primary care using all the methods of science and the array of settings in which primary care is delivered. At present, no adequate infrastructure exists that is designed to undergird an enduring primary care research enterprise. In this committee's view, it is unlikely that primary care can be grounded in an adequate science base unless such infrastructures are created.
The untapped opportunities in primary care research leave us ignorant about why some people get sick while others stay well and why some people recover from their illnesses and others do not. Primary care research can determine the transition of signs and symptoms and vague concerns into clinically more significant diseases and diagnoses so that prognostication can be improved and the needs of newly forming integrated delivery systems and the patients that they serve can be met. The most urgent need, however, is not for a particular investigation but the building of the nation's capacity to investigate multiple primary care questions. In other words, the overriding goal must be to establish a viable primary care research infrastructure.
Key elements of such an infrastructure have been defined. Among them are
- a designated lead agency at the federal level that would be held accountable for advancing primary care research;
- national health and health care utilization surveys and databases that capture the relevant aspects of and data on primary care;
- primary care research laboratories, such as practice-based research networks that link primary care practitioners with those who carry out scientific investigations;
- appropriate data standards and classification systems for primary care;
- training programs for primary care clinician-scientists; and
- stable career ladders for primary care researchers.
The committee found exemplary efforts in each of these areas. Some are well known, such as the large population surveys of the National Center for Health Statistics (NCHS) or the Agency for Health Care Policy and Research (AHCPR) in the Department of Health and Human Services (DHHS). Similarly, the international classification system for primary care (the International Classification of Primary Care, or ICPC) is widely known in certain circles (especially abroad) and clearly opens the door to the episode-oriented epidemiology critical to capturing the phenomena of primary care.2 Other advances are not widely
recognized or are too new to be broadly known; the idea of primary care laboratories or primary care practice-based networks for research (discussed below) is a case in point. Yet others are in a developmental stage; the lead agency for primary care research falls into this category.
In many ways, apart from long-standing federal surveys, what has been achieved so far has been largely through a patchwork of research efforts and the successes of occasional champions, relying on budgets that seem trivial when compared to the nation's other research commitments relevant to health, such as the biomedical research institutes. The future of primary care will be strengthened if genuine capacity for conducting primary care research is clearly established. The following subsections discuss in more detail specific needs that must be met to set a primary care research infrastructure solidly in place; where relevant, the committee's recommendations are given.
Federal Leadership and Support for Primary Care Research
The nation might be said to have either a surfeit of agencies conducting or supporting what they regard as primary care research or essentially no such capacity. The stand one might take on this depends in part on the breadth and content of one's concept of primary care research. Certainly if the focus is on research that is carried out in primary care settings to answer epidemiological, clinical, organizational, or other questions about primary care needs and health care delivery, then capacity and output to date have been low. That is essentially the conclusion of this committee. To address this gap, the committee reached consensus on the proposition that substantially greater emphasis, focus, and support for primary care research is needed at the federal level.
Recommendation 8.1 Federal Support for Primary Care Research
The committee recommends that (a) the Department of Health and Human Services identify a lead agency for primary care research and (b) the Congress of the United States appropriate funds for this agency in an amount adequate both to build the infrastructure required to conduct primary care research and fund high-priority research projects.
A lead agency is necessary for two main reasons. First, primary care research should not be a crazy quilt of independent research efforts, none working with another, none building on previous or ongoing work. Thus, the coordinating
and ''bully pulpit" functions of a lead agency will be very important in developing synergistic programs of research and in raising the visibility of primary care research. Second, this is a period of unprecedented belt tightening at the federal level, and the nation cannot afford to squander any resources devoted to primary care research. A lead agency is thought to be a useful vehicle for shepherding the nation's scarce research dollars in this area and for making the most efficient use possible of the resources that are available.
Placement and Role of a Lead Agency
The committee did not take a stand on precisely what unit within the Department of Health and Human Services (DHHS) might be assigned the lead, as it judged that the best decision could be made by the Secretary of DHHS in light of evolving organizational change within the department. The committee recognizes, however, the importance of AHCPR as the only federal agency explicitly authorized to investigate primary care as it is conceived of by this committee, and the agency has mounted an important effort to do just that in the past two years or so. Furthermore, in mid-1995 the reorganization plan for the agency (Federal Register, 1995) included a "Center for Primary Care Research." It was assigned the responsibility of conducting and supporting projects in the following five areas: (1) primary care settings and systems; (2) rural health care services and systems; (3) care for special populations; (4) effectiveness of education, supply, and distribution of the health care workforce; and (5) international activities. To the committee's knowledge, no other element of DHHS has this charge or responsibility.
As this report was being prepared, the Institute of Medicine issued a report on the health services research workforce, and it noted that the leading sponsor of broad-based health services research in this country is AHCPR (IOM, 1995). The primary care committee observed that many of the issues addressed by health services researchers for the past quarter-century, many of the settings in which it is conducted, and many of the methodological and statistical approaches employed by health services researchers are relevant to primary care research as envisioned here.
The responsibilities of such a lead agency could be to conduct, oversee, and coordinate activities relating to a broad agenda of primary care research (see below). In addition, it could assist or advise other agencies on matters relating to the supply of primary care clinicians or of appropriately trained researchers; for example, a lead agency for primary care research could work with the Bureau of Health Professions (BHP) in the Health Resources and Services Administration (HRSA) in that agency's efforts to monitor the supply of physicians, nurses, and other health professionals. Furthermore, a lead agency could support research training programs as well as offer technical and methodologic assistance to other federal, state, or local agencies wishing to pursue primary care research on their
own. A lead agency would be in a strategic position to synthesize and disseminate information from the full array of primary care research projects being carried on in the nation (whether it is actively funding them or not). A final, important responsibility of such an agency would be to build an understanding of the need for such research and a constituency that will demand, use, and act on the results of such projects.
Collaborative Responsibilities of a Lead Agency
A lead federal agency for primary care research should interact and collaborate with both other elements of DHHS and other federal departments. An agency within the Public Health Service (PHS), such as AHCPR, could develop cooperative arrangements and communication avenues with other PHS agencies. These include the Centers for Disease Control and Prevention (CDC), offices and bureaus in HRSA (such as BHP and the Maternal and Child Health Bureau), and the National Institutes of Health (especially those institutes that carry out clinical investigations about primary care concerns or support clinical trials in ambulatory settings). Collaboration with the CDC might be especially important, given that agency's responsibilities in the area of prevention.
In addition, a lead agency for primary care research (if placed in the PHS) would need to develop collaborative relationships with several other DHHS agencies. Primary among these is the Health Care Financing Administration (HCFA), given its responsibilities for the Medicare and Medicaid programs and its own health services and policy research agenda. Other elements of the department also have interests that intersect with primary care research because of their focus on ensuring high-quality services to specific populations; among these are the Administration on Aging and the Administration on Children and Families.
A formally designated lead agency could also forge links with two other departments with significant primary care concerns and responsibilities for large numbers of individuals across the age span—namely, the Department of Veterans Affairs (VA) and the Department of Defense (DOD). Research efforts sponsored, for example, by the VA's Health Services Research and Development program or by DOD/Health Affairs (which cuts across all the armed services) could present important opportunities for joint research efforts and interaction, especially to the extent to which VA and DOD efforts are carried out in primary care settings.
Many entities in the private sector underwrite various types of primary care research or programs to build capacity for such work. For example, the Robert Wood Johnson (RWJ) Clinical Scholars program has produced, through its more than 20-year history, a cadre of primary care clinicians capable of independent health services and policy research. Other RWJ programs have involved Generalist Academic Fellows in pediatrics and medicine. The Pew Charitable Trusts have for 10 years also supported training in health services research, and many of its Fellows come from or conduct research in primary care disciplines. The
community development orientation of The W. K. Kellogg Foundation has also provided a context for innovative demonstration projects that relate to the delivery of primary care services. In addition, the committee learned of both medical and nursing schools that support primary care faculty to some extent to carry out research in this field.
Finally, as discussed more fully below, more than 30 practice-based research networks are either operating or serving as laboratories for various kinds of primary care research (see Recommendation 8.3). The committee judged that collaborative public-private partnerships are the wave of the future (see also Recommendation 9.1 in the following chapter). Thus, it believed that, in all these cases, a lead agency could better foster mutually productive relationships than could multiple agencies operating independently.
Data Sources and Needs
Need for a Primary Care Database
Research often demands complex primary data collection, particularly when specific hypotheses are to be tested using experimental or quasi-experimental designs. By and large, the details of such data sources and methods for collecting or analyzing such information must be specific to the research project at hand. Although the committee did not explore the issue of primary data collection directly, it was cognizant of the fact that much useful work can be done with analyses of secondary, survey, or administrative data, providing that the data are reliable, valid, accessible, and current. Such information offers a means by which the relevant phenomena of primary care can be captured to answer such questions as "What is the great majority of health care needs to which primary care responds?" "Who is delivering that care today?" "In what settings?'' and "At what cost?" Thus, the committee looked into the characteristics of major surveys and databases that might be seen as part of the infrastructure for primary care research, giving particular attention to whether they collect and array data in ways that would permit analyses to be done on episodes of care.
Existing Data Sources Relevant to Primary Care Research
The federal government conducts a great variety of surveys and oversees an array of databases that have relevance for primary care and primary care research.3 These are the responsibility of several departments in the executive
For more detail on the nature and breadth of federal health data sources, see CBASSE/IOM, 1992, especially Appendix C, and the sources cited there. As this report was being prepared, DHHS was designing a major restructuring and consolidation of core DHHS surveys, as a means of reducing duplication and increasing survey efficiency, meeting a broad set of user needs, filling existing data gaps, and improving the efficiency of the survey enterprise. As the details of these proposals were not available to the committee, it was not able to take them further into account for this report. It is clear, however, that the committee's ideas about a primary care survey and database (discussed in Recommendation 8.2) are consistent with the general thrust of the DHHS plans.
branch. The great majority of these activities reside, of course, in DHHS. A selected few, briefly described below, are of special importance for primary care research; these include the programs of NCHS (which is the lead agency for the "production, analysis, and dissemination of general-purpose health statistics" [CBASSE/IOM, 1992, p. 129] and is a part of the CDC), those of AHCPR (the lead agency for health services research), and those of HCFA (which administers the Medicare program and the federal portion of the Medicaid program).
NCHS conducts many general and special purpose surveys. The best known are the National Health Interview Survey (NHIS) and the National Health and Nutrition Examination Survey (NHANES). The NHIS is a cross-sectional household interview sample survey providing information on the general health status of the civilian, noninstitutionalized population; it yields national estimates of the incidence of acute illnesses and injuries, the prevalence of chronic conditions and impairments, the use of health care services, and various other health-related topics. This annual survey comprises both a core set of questions (on measurement of illness and injuries, days of disability, limitations of activities, use of health care, perceived health status, and socioeconomic characteristics) and a variety of items (in supplements) that change from year to year depending on the health issues of the day.
The NHANES evolved from the Health Examination Survey first launched in 1959. It involves direct standardized physical examinations, clinical (e.g., visual acuity) and laboratory (e.g., biochemical and hematologic) tests, measurements of nutritional status, and interviews. The aim is to obtain information on "undiagnosed" and "nonmanifest" diseases and the prevalence of defined diseases or conditions as well as to gather normative health-related data.
Since 1973, the National Ambulatory Medical Care Survey (NAMCS), a national sample survey, has gathered and disseminated information on patient visits to physicians' offices. It excludes services provided by nonphysician personnel and telephone contacts. NAMCS collects information on characteristics of patients, diagnoses and symptoms, diagnostic and therapeutic interventions, and characteristics of the physician and payment source. Only since the establishment of the National Hospital Ambulatory Medical Care (NHAMC) survey in 1991 has similar information been collected on health care provided by hospital emergency and outpatient departments.
AHCPR administers the National Medical Expenditures Survey (NMES), which is intended to provide national estimates of the use of and expenditures on
health care services and the extent of health insurance coverage in the nation. The three major components of this survey (a household survey, an institutional population component, and a survey of American Indians and Alaska Natives) yield population-based estimates of health and functional status, insurance coverage and health care utilization, outlays, sources of payment, and various other socioeconomic data on respondents. A long-term care supplement provides information on persons with functional disabilities and impairments and their use of formal home and community-based services. This survey, although comprehensive with respect to these topics, is conducted only infrequently, and data are not readily accessible; it was undergoing considerable revision in terms of sampling design and similar matters as this report was being prepared.
A final survey of great potential for primary care focuses on the Medicare beneficiary population. Administered by HCFA, the Medicare Current Beneficiary Survey (MCBS) is intended to be an ongoing multipurpose interview survey focused on health care use and expenditures; it also includes health and functional status and various socioeconomic and demographic characteristics, including family support. This survey, because it is of a representative panel of Medicare beneficiaries over time, can provide data sets appropriate for either longitudinal or cross-sectional studies.
Although all the above are critical elements of the nation's infrastructure for health statistics, and although all yield some information pertinent to primary care research, they are not (either individually or collectively) completely satisfactory as a base for analyzing the range of issues that this committee believes ought to be included in the primary care research agenda. For instance, with the exception of the MCBS, they are not person-specific and do not yield longitudinal data on particular individuals; this means that episodes of care for specific people and their particular conditions cannot be created or analyzed from these files. In addition, they do not include information on nonphysician primary care clinicians at the level of detail necessary to understand who is providing what kinds of services to which patients for which problems; thus, they cannot illuminate the concept of the primary care team.
The committee spent some time discussing the strengths and limitations of various large-scale data sources of this type, with specific attention to the national surveys conducted by DHHS agencies. In general, none of these surveys was considered wholly adequate for the purpose of providing information to answer a broad set of primary care research questions, and indeed even collectively they would not suffice. To address this need, therefore, the committee agreed on a recommendation concerning a primary-care-specific database that would be created from a periodic national sample survey.
Recommendation 8.2 National Database and Primary Care Data Set
The committee recommends that the Department of Health and Human
Services support the development of and provide ongoing support for a national database (based on a sample survey) that reflects the majority of health care needs in the United States and includes a uniform primary care data set based on episodes of care. This national survey should capture data on the entire U.S. population, regardless of insurance status.
Episodes of Care
Defining episodes of care. Key elements of the committee's definition of primary care involve continuity and coordination of care. Research in this area must, therefore, be able to track the care for specific conditions through time and across clinicians and settings. This condition in turn requires the ability to create what analysts term "episodes of care." An episode of care has been defined as "all care for a given spell of illness, for a specific injury, or for a particular chronic illness" (Lohr et al., 1986, p. S9); this is similar in concept to the definition advanced in Chapter 4 and its appendix that an episode refers to a problem or illness during the time from its first presentation to a clinician until completion of the last encounter for that problem.
Ideally, one would prefer to study episodes of illness and of disease ; for economic analyses at least, an episode of illness has been characterized as the "natural unit of decisionmaking" (Newhouse and the Insurance Experiment Group, 1993, p. 80). An entire episode of disease is not by and large a measurable concept, because the beginning date of an acute or (especially) a chronic problem can rarely if ever be known, especially for conditions with insidious onset or lack of symptoms until the disease is well advanced. The exceptions may be episodes starting with an acute traumatic injury and well-child care starting with the birth (although the latter is not typically regarded as illness). In addition, long-term chronic disorders that have periods of remission and flare-up add to the conceptual complexity of an episode of disease.
For these reasons, most episode-based research focuses on episodes of care. These typically date to an office visit, emergency room visit, hospitalization, or the like that appears to be the first identifiable service for a symptom or condition; they end, depending on the analysis in question, at death, at the end of traceable services for that problem, or at the end of a year.4
The content of an episode of care can be extremely complex; even for ambulatory care, it can include visits to various clinicians (or visits by various clinicians to a home or other setting), laboratory tests (and ideally their results), diagnostic procedures (and results), and both inpatient and outpatient treatments of all sorts, including prescription (and, ideally, nonprescription) medications. This characterization of episodes of care is not easy to realize on a routine basis, however, and in many circumstances an "episode" may relate only to visits linked over time.
Reasons for using episodes of care. The analytic rationale for using episodes involves the ability (a) to describe better the full set of services directed at a particular clinical problem or provided to a specific individual for preventive or for diagnostic and therapeutic reasons and (b) to capture more fully the cost and quality consequences of the patterns of care. In addition, using episodes of care permits better modeling of care decisions by both patients (or family members) and clinicians than would be possible simply with counts of physician or clinic visits.
Barriers to creating and using episodes of care. By and large, current databases (such as the survey files described earlier) cannot provide the types and levels of information required to create, let alone analyze, episodes of care. The ability to examine complete episodes of care that may extend over time and space is a critical feature of a "primary-care-informative" data set that most existing databases do not have.
None of the current data sets adequately overcome the limits of their cross-sectional design. For example, even though some surveys (such as the NHIS) ask respondents to give their answers about events in terms of a recent time frame, the files are not (and cannot be) constructed in a way that would permit care to be viewed as a longitudinal series of visits and activities that occur over time and in different parts of the health care system. To give another illustration, past NMES databases link events for individual patients over time and thus in theory might have been very useful in primary care research, but the 1987 NMES was not designed for this purpose and may not contain the necessary level of detailed data
on physicians seen and services provided (as contrasted with economic information on charges and reimbursements or out-of-pocket expenditures). Finally, the likely requirement to maintain the confidentiality of respondents' identity in survey data amassed with public funds can clearly hamper any linkage of information over time.
Somewhat the same problems plague administrative insurance claim files (such as the Medicare databases). Medicare files—which many regard as the best source of condition-specific utilization information that the nation has today—do not, for example, cover prescription medications and do not code laboratory tests or other services at a sufficient level of detail. (Obviously, these files yield no information on the general population of nonelderly individuals.)
Good insurance files for private sector plans that cover a generous benefit package (explicitly, those that cover preventive services and prescription medications in addition to traditional outpatient and inpatient services) can be used to create episodes of care, but doing so requires considerable clinical, analytic, and computer expertise to develop the "rules" by which various different kinds of services are linked correctly into diagnosis- or problem-specific episodes over time. Furthermore, such files may or may not be available for research purposes, depending on who owns them, although clearly such organizations can carry out internal research related to primary care. By and large, patient records and data files maintained by health maintenance organizations, physician networks, or integrated delivery systems today cannot provide, in any routine way, the episode-based information that would be needed to measure primary care. In any case, it would be unusual for specialized episode-of-care information to be made available to outside investigators; but as with insurance files, these managed care organizations can and do conduct their own studies on primary care issues (which might be episode based).
It is quite possible to create episodes of care using algorithms that are based on diagnoses; indeed, this is the preferred approach. This was done in early work in the Indian Health Service (Paul Nutting, University of Colorado Health Sciences Center, personal communication, September 30, 1995), for example, and much of the analyses done in the RAND Health Insurance Experiment were based on complex diagnosis-specific episodes created from the research-oriented insurance claim files of that study (see, for example, Keeler et al., 1982; Lohr et al., 1986). Today, unfortunately, coding of diagnoses may often be related less to the reality of patient illness than to the complexities of reimbursement rules established by third-party payers (IOM, 1994b) or to the desire to circumvent stringent physician profiling programs of managed care or accreditation organizations.
Incompatibility of concepts. Apart from the conceptual and practical issues about episodes of care discussed above, the committee acknowledges the likely difficulty of operationalizing several parts of its definition for survey or database
purposes. To illustrate the point: In the data exercise conducted for the committee, the notion of "practicing in a family and community context" proved especially problematic. Although, clearly, reliable survey questions can be devised to tap this domain, a nontrivial amount of methodologic work might need to be done to capture the concept adequately in terms that lay respondents would understand.
Getting at these concepts through present-day administrative databases or existing surveys is almost certainly not possible. One might try to use diagnostic or reason-for-visit codes of "family problems" or similar nomenclature as an indicator measure of family context. In the data exercise conducted for the committee, however, this diagnostic cluster appeared too infrequently to be useful. Another alternative might be to examine patterns of diagnosis-specific services that are rendered to several members of one family. An example might be the constellations of (ostensibly related) services for one parent who is terminally ill with cancer, counseling for that individual's young children who have behavioral problems at school or are visibly depressed, and respite services for the spouse. The methodologic challenges of creating such analytic units are extreme, however. Moreover, in general the committee believed that practicing in a family and community context implies a great deal more than simply providing related (or unrelated) care to more than one family member. In addition, the above analytic tactics obviously do not adequately address the idea of "community."
Finally, none of the current data sets adequately addresses the problem of properly labeling the events of primary care. This requires attention to both the patient's multiple reasons for visits and the physician's appraisal of the problem(s) as it (or they) evolve over multiple visits. Classification systems now mainly used outside the United States (such as the ICPC) may, however, help overcome this problem if they become more widely used here (Lamberts et al., 1993). The point is taken up again in the discussion of data collection standards below.
Structure of a Primary-Care-Oriented Survey
The committee did not discuss in any depth the details of a new or modified survey that would satisfy the goals set out in Recommendation 8.2, nor did it come to consensus on what federal entity within DHHS might conduct such a survey on a routine basis. It did, however, believe that such a survey should follow a probability sample of individuals of all ages over time and that it should focus on core elements of the primary care definition.
One possibility might be for such a survey to be fielded independently of any of the current NCHS surveys but to use the same or a variant of the person-specific (rather than provider-specific) sampling frames used in, for instance, the NHIS or NHANES. Yet another option might be for the next NMES (now referred to as the Medical Expenditures Panel Survey, or MEPS) to be expanded or for a supplement to be fielded that would target significant primary care questions.
Toward A National Health Care Survey (CBASSE/IOM, 1992) had recommended that NCHS develop a new survey that would address current limitations of certain NCHS surveys and improve their ability to measure the longitudinal dimensions of care. The aim was to overcome one major drawback, namely, that the starting point is a sampling of events rather than a sampling of patients. That committee recommended that person-level data be collected "on health care received by individuals over time and over the entire progression of an episode of illness" (p. 4). Moving in this direction would clearly go a long way toward reaching the goal of the present committee with respect to a primary-care-oriented national database.
Regardless of the mechanics,5 the committee wishes to go on record as favoring large-scale, episode-oriented morbidity studies in the population that can provide the databases needed to enable the nation to assess how well the concepts in the IOM definition are being achieved. In the research arena the committee has emphasized surveys or other types of data collection strategies that would feed into databases that could be tapped for various types of primary care studies. In addition, the committee calls attention to the opportunities offered by practice-based research networks to be an important source of such survey information (see the discussion for Recommendation 8.3 below). Because such information would come from the front lines of primary care practice, the resulting database would comprise clinically important information for research, policy, and practice applications.
The committee also holds that the development of computer-based patient records ought to be a high priority as well, consistent with the recommendations of two earlier IOM committees (IOM, 1991a, 1994b). In this way, it may be possible in the future to have a data collection structure that permits data to be organized into episodes of care in sufficient clinical and demographic detail to measure the components of primary care far more appropriately than is true today. In particular, the committee would urge managed care organizations to adapt their information systems in such a way that they can measure episodes of care for their enrollees within the conceptual framework of the definition of primary care given in Chapter 2.
Primary Care Research in Practice-Based Research Networks
Several barriers to primary care research need to be overcome. Among them are "a lack of a critical mass of researchers, competing demands faced by investigators, lack of a research culture, and difficulties in operationalizing practice-based wisdom into researchable theories, measures, and appropriate study designs" (Stange, 1996). The shortage of funding for research in primary care, which Stange attributes in part to the typical "categorical" nature of much research support and which is not in keeping with the generalist underpinnings of primary care, is a major obstacle.
Another major unmet need lies in the area of simultaneously providing sophisticated methodologic training for a relatively small cadre of primary care researchers while also opening up opportunities for large numbers of clinicians to participate in such research (Stange, 1996). Nerenz (1996) also discusses the strengths and limitations of primary care research within integrated health systems. Multidisciplinary collaboration and support for such work must be increased,6 through mechanisms such as primary care research centers and practice-based research networks.
According to Nutting (1996), studying the relevant phenomena of primary care presents a logistical challenge that might be satisfactorily addressed through the use of practice-based research networks. Analogous to the networks of major tertiary care centers that conduct the great bulk of the nation's basic biomedical research, primary care networks serve as laboratories by which health care events and the health status of many patients might be studied in "real world" settings.
Although less well known in this country than elsewhere, practice-based research networks have operated in other countries for years, often with considerable support from central governments. Niebauer and Nutting (1994) reported that at least 28 such networks now operate in North America. Among them are networks associated chiefly with family medicine that date to the mid-1970s, including the Dartmouth Primary Care Cooperative Information Project (or Dartmouth COOP), the Ambulatory Sentinel Practice Network (ASPN), the Michigan Research Network, the Minnesota Academy of Family Practice Research Panel and the Wisconsin Research Network (WREN). The American Academy of Pediatrics practice network has also been in operation since the 1980s.
Nutting (1996) describes practice-based research networks as having the following four characteristics:
- Capturing health and health care events relating to community-based patient populations;
- Providing access to the full range of practice experiences of all primary care clinicians;
- Focusing activities on research questions that are relevant to actual practice, using sophisticated research designs and statistical methods; and
- Involving in a systematic way the networks' own clinicians in defining the research issues, designing the project, and interpreting the study results.
The committee sees practice-based research networks as a significant underpinning for studies in primary care, noting not only their attractiveness conceptually but the growing recognition of their value as reflected in the rise in the number of such entities in recent years. For this reason, the committee reached consensus on a recommendation that these enterprises should receive high-priority attention and funding to carry out the variety of studies that will contribute to the science base of the future for primary care.
Recommendation 8.3 Research in Practice-Based Primary Care Research Networks
The committee recommends that the Department of Health and Human Services provide adequate and stable financial support to practice-based primary care research networks.
Using Practice-Based Research Networks
Some commentators raise the question of directing such support to primary care research networks, believing that such groups can compete satisfactorily in the traditional investigator-initiated research mode. In the committee's view, however, these types of research networks have been successful in the past and today offer the most promising infrastructural development it could find to support better science in primary care.
Primary care research is not being defined as research solely about problems that exist in primary care; as noted earlier, it is not to be conducted in a vacuum. Rather, the committee is acting on its understanding of the ingredients of an effective scientific enterprise: Scientists have to get the questions straight and observe the relevant phenomena, so that their investigations can be done with reliable estimates of error and bias. Primary care practices, and the research networks that link them, are the sources of the right questions, and they offer an efficient mechanism of relating those questions to the appropriate events that need to be studied.
According to Kerr White (Charlottesville, Va., personal communication, December 21, 1995):
Asking the "right" questions is at the heart of all research. Asking "important" and "researchable" questions is also an essential element of every investigator's basic preparation. Describing the distributions of diseases, conditions, services, interventions, and costs is of great importance. Equally if not more important, however, should be deep concern on the part of the primary care establishment, especially its academic components, with adding to the essential fund of knowledge bearing on the origins of diseases and ultimately their prevention or amelioration. The enormous strengths of biomedical research during the past half century have been its outstanding contributions to our understanding of disease mechanisms and processes and the development of a cornucopia of efficacious interventions. The primary care fraternity in recent decades has contributed precious little to medical science. … If primary care is to take its rightful place in medicine and the healing sciences, then it must contribute to fundamental medical knowledge, not just to knowledge about services, education, and training, and about epiphenomena bearing on its own ministrations.
Funding for Research Through Practice-Based Networks
A major concern about these research networks—which in reality generalize to the entire primary care research enterprise—is predictable funding. Stable financial support is necessary for two principal reasons. First, it enables investigators with ambitious and innovative projects to begin and complete them and to disseminate their results. Second, it sends a message to researchers and clinicians (i.e., potential investigators or participants in research projects) that they can have reasonable expectations of a career in primary care research.
The committee did not, in the end, establish a target figure for research funding in primary care. After protracted debate on the point, it did agree on the following propositions. First, the level of funding ought to be proportionate to the nation's outlays on primary health care. The bulk of the medical enterprise in this country lies in primary care, but the bulk of the research enterprise lies elsewhere. The disproportion between that investment (i.e., in basic biomedical research and clinical investigation) and the resources directed at primary care research is extreme, and the committee believed that some redress of this imbalance is in order.
Second, the committee estimated that federal investments in primary care research today total between $15 million and $20 million annually, depending on what is included. This level is one that the committee regards as absurdly inadequate to the task. Although arriving at a realistic figure for federal funding of primary care research was not possible, the committee generally believes that levels of support four to five times the present investment were not unreasonable. An amount of this magnitude, the committee judged, would enable the nation to support useful research projects and, at the same time, begin to build the infrastructure required to carry the primary care research enterprise forward into the next century.
As noted in Chapter 7, the committee recommends an all-payer approach to support education and training in primary care (see Recommendation 7.5). The committee believes that this strategy can and ought to be used to lend some long-term stability to the funding for primary care research as well. One tactic might be to direct a small, specified percentage of whatever sums are raised through an all-payer program for education and training to the "lead agency" specified in Recommendation 8.1 above, for the explicit purpose of broad-scale support of research or more targeted funding of primary care research networks or other centers that may combine training and research.
Finally, although Recommendation 8.1 is directed at DHHS, the committee wishes to go on record as urging the nation's premier health care foundations to promote an even greater level of research activity in primary care. The general cutbacks in federal research dollars are coming just at the time that the need for increased investment in primary care research is becoming acute. Thus, although foundations have played a role in this arena for some years, in the committee's view it is even more important for them to do so now. The committee would encourage foundations that have not been deeply involved with primary care issues to enter the field, and it sees this as a major area in which foundations now being created from former health care provider and payer organizations might have a considerable interest.
Standards for Data Collection
The committee was very aware of the significant problems that occur in using existing data for primary care research purposes. These include consistency of definitions of clinical, health services, epidemiologic, and demographic variables; coding of diagnoses and procedures; and similar data standards problems. These matters extend well beyond primary care; for example, a similar plea for improved coding and definition of minimum data sets was put forward by the IOM committee on emergency medical services for children (IOM, 1993b) and by a different committee concerned with the use, disclosure, and privacy of health data (IOM, 1994b). In discussing the necessary "upgrades" in the primary care research infrastructure, however, the committee concluded that specific attention to standards for data collection that could be promulgated by appropriate authorities in both the public and the private sectors is necessary.
Recommendation 8.4 Data Standards
The committee recommends that the federal government foster the development of standards for data collection that will ensure the consistency of data elements and definitions of terms, improve coding, permit analysis of episodes of care, and reflect the content of primary care.
The issues represented by this recommendation are broad indeed. The committee has commented elsewhere in Chapter 4 and earlier in this chapter about the desirability of conducting research in terms of episodes of care. Doing so, however, calls for better classification systems for reasons for visit (or encounter), diagnoses, and services. This in turn requires that all those responsible for developing databases or recording and reporting health data work from a consistent understanding of desirable (or at least necessary) data elements and of how those data elements are to be defined and described. The difficulties that this set of requirements presents should not be underestimated. As just noted, at least two other IOM committees in recent years have directed a considerable amount of attention to data and data systems (IOM, 1993b; IOM, 1994b), and readers are directed to those reports for more detailed discussion of these issues.
Coding of clinical conditions, diagnoses, symptoms, and complaints as well as services rendered is especially problematic, particularly if the full range of primary care as envisioned by this committee is to be adequately taken into account and reflected appropriately in research data sets. Also likely to be questionable is the information typically recorded on what types of practitioners (e.g., physicians, nurse practitioners, or physician assistants) may have actually provided the services in question. Lacking reliable approaches for defining and coding such information, certain types of primary care research may face considerable methodologic challenges.
A final comment is that the recommendation calls for the federal government to develop standards for data collection. Among the issues that such an effort might address are formal, quantitative or qualitative standards involving reliability, validity, and practicality of data collection instruments and methods. One background guide for such effort might be the recently published criteria from the Medical Outcomes Trust, by which it evaluates standardized measurement instruments in the area of health outcomes (Perrin, 1995). These criteria include specific elements related to: the conceptual and measurement model; reliability (both internal consistency and reproducibility [test-retest and inter-observer or inter-interviewer reproducibility]); validity (content, construct, and criterion); responsiveness (ability of the instrument to detect change); interpretability (the degree to which one can assign qualitative meaning to quantitative scores); respondent and administrative burden; alternative forms of an index measurement instrument; and cultural and language adaptations of an index instrument (Scientific Advisory Committee, 1995). This listing alone makes clear the challenges of adequately meeting the committee's aspirations in this area.
Priority Areas For Primary Care Research
As noted above, many issues confronting investigators in primary care are similar to those addressed by health services researchers. The committee wished
to highlight certain topics, however, as they relate to the committee's conceptualization of primary care (in Chapter 2) and its view of the settings within which primary care research ought to be conducted. Thus, this section briefly discusses the links that must be forged with elements of the infrastructure discussed above and with other types of research. It then offers a selected set of topics that the committee judged, drawing on the outcomes of its research workshop and its own expertise, warranted early or high-priority attention.
Links to Infrastructure Capabilities
As should be clear by this point, the committee concluded that the scientific underpinnings for primary care are of uneven quality. The mismatch between the scope of, expectations for, and potential of primary care (on the one hand) and the research base that documents those characteristics of primary care (on the other) is considerable. The recommendations offered above for strengthening the infrastructure, visibility, priority, and funding for primary care research are means by which the committee hopes that this situation will be ameliorated. Nonetheless, the committee recognized that resources for primary care research are always likely to be scarce (at least relative to desirable levels) and that priorities need to be set on researchable questions that warrant near-term attention. One way of organizing the myriad suggestions and opportunities for primary care research is to focus on the infrastructure elements discussed already in this chapter.
Having a lead agency is critical to operationalizing the research agenda from the perspective of federal funding. A lead agency might also be especially well placed to foster cross-cutting research projects that link ''basic" primary care research with other research areas, such as methodologic investigations of outcomes or patient satisfaction measures or studies of the impact of family support systems or cultural competency on the need for long-term care services.
A national, survey-based data set that captures episodes of care contributes to understanding the epidemiology of health care needs (e.g., incidence and prevalence of disease; utilization of services). Such data files also provide critical information about the access to care that all members of the population (however defined) actually have and about the integration (i.e., continuity and coordination) of health care services. Improved data collection and coding standards, especially if linked to primary care research supported by a lead agency, will reinforce efforts to explore issues relating to the large majority of health care needs. Finally, practice-based research networks are intended to provide the venues in which all of the core elements might be studied, but they would be in a particularly good position to investigate issues of access to care, coordination and continuity, sustained partnerships, accountability, and family and community context; a special point about practice-based networks may be their ability to investigate how various parts of the nation differ in these respects.
An underlying issue for this committee is the recognition that research in this area requires that health be understood not just in relation to traditional biomedical models, but also in terms of physical and cognitive functioning, emotional well-being, and changing states of health. The concepts of functioning and well-being are multidimensional (Inui, 1996), and this very "messiness" makes primary care research both challenging and rewarding in terms of what can be learned about effective health care. It also calls attention to other important linkages, particularly with health services and clinical research and between physical and mental health.
Links to Health Services Research
The scope of primary care research includes or interacts with other major areas of health services research (IOM, 1995). Among these are outcomes and effectiveness research; quality assessment and improvement; development and dissemination of clinical practice guidelines; a wide array of methodologic questions, including refinement of instruments to measure health-related quality of life, patient utilities and preference weights, and improved methods for severity and risk adjustment; organization and financing of health care delivery and the general area of health economics; health professions workforce (e.g., effective education and training programs, supply and demand modeling); sociological issues (e.g., the role of social support and self-efficacy); and information systems (e.g., uses of computer-based patient records and clinical applications of telemedicine and telecommunications technologies).
Put another way, the above-mentioned topics are not unique to primary care. Rather, they are applicable in many instances to primary care; conversely, research in the primary care arena is likely to contribute to advances in the knowledge base on these issues. For example, the involvement of ambulatory patients is likely to provide insights about patient or practitioner satisfaction or outcomes of care quite different from those emanating from research in the inpatient or long-term care setting.
Links Between Primary Care Research and Clinical Trials
Almost all research and teaching in the United States today are conducted in tertiary care centers. The knowledge derived there, especially from classic randomized clinical trials (RCTs), does not easily apply to primary care. Generalization to the patients seen in primary care practices is difficult because of exclusion criteria commonly set for study populations. For example, RCTs often exclude patients with comorbidities and indistinct conditions, individuals of certain ages (especially the elderly and healthy children) or gender (until recently, especially
women of childbearing age), and hard-to-reach or non-English-speaking groups. Furthermore, because of cost and time constraints RCTs must use intermediate (often anatomic or physiologic) end points rather than broad measures of health status, functioning, and health-related quality of life.
These limitations do not fit the model of primary care practice and thus are drawbacks from the point of view of primary care research. Patients seen in primary care settings can present with multiple diagnoses, puzzling complaints, and unacknowledged disorders (e.g., mental and emotional trauma). This comorbidity is part and parcel of the primary care enterprise, and it may or may not be well reflected in classic RCT investigations.7
Links between Physical and Mental Health
A particularly important element of primary care involves the fact that emotional and physical problems tend to be intertwined, such that one can only be understood in its relationship to the other and to the context of a patient's life. In a paper prepared for this committee (see Appendix D of this report), deGruy makes the following point:
Systems of care that force the separation of "mental" from "physical" problems consign the clinicians in each arm of this dichotomy to a misconceived and incomplete clinical reality that produces duplication of effort, undermines comprehensiveness of care, hamstrings clinicians with incomplete data, and ensures that the patient cannot be completely understood.
This committee agrees: Those who can and do take responsibility for the quality of patient care in a primary care setting (indeed, in any setting) must never lose sight of this inextricable, inevitable relationship between the physical and mental domains of health. The lesson for researchers is that they must themselves undertake to examine these domains of health in tandem, especially in projects that relate to continuity and coordination, accountability for quality of care, and family and community contexts.
Primary Care and Specialist Physicians
The committee has not, to this point in this chapter, singled out the concept of "primary care clinician" as an appropriate target of research (recalling that the term clinician in this context refers to physicians, nurse practitioners, and physician assistants [see Chapter 2]). On numerous occasions throughout the study, however, two questions arose: Who today actually delivers primary care, and to what effect on patient outcomes? Of particular concern was the notion that in the future even more physicians (than is true at present) who are trained and practicing in various specialties and subspecialties will be providing primary care services. In the committee's view, this phenomenon deserves explicit attention as part of a broad primary care research agenda. For that reason, it states the fifth recommendation of this chapter.
Recommendation 8.5 Study of Specialist Provision of Primary Care
The committee recommends that the appropriate federal agencies and private foundations commission studies of (a) the extent to which primary care, as defined by the IOM, is delivered by physician specialists and subspecialists, (b) the impact of such care delivery on primary care workforce requirements, and (c) the effects of these patterns of health care delivery or such care on the costs and quality of and access to health care.
The committee reached this recommendation after reflecting on several trends in health care today, especially the accelerating growth of various types of managed care organizations. First, it has become clear that a good deal of primary care is delivered today by specialists and subspecialists, particularly those in internal medicine and pediatrics. Second, many of these practitioners will likely try to raise the proportion of primary care in their practices, as a result of specialist oversupply (see Chapter 6) and the increasing difficulty of making a career solely in specialty practice. Third, no one can say with certainty what net effects these overlaps in primary and specialty care have on costs and quality (IOM, 1996a).
These issues intersect with some already alluded to, including those relating to access to care, the idea of a sustained partnership between patients and clinicians, and accountability for high-quality health care. The committee judged, however, that the basic phenomenon was sufficiently troublesome to warrant a very high ranking in any primary care research agenda. The reasons are several: problems of quality of care if physicians practice outside their usual areas of competence; questions of the adequacy of "retraining" programs for subspecialists (see Chapter 7); and ramifications for overall physician workforce supply and demand modeling and projections. Also of concern is the appropriateness, from
either a cost or a quality standpoint, of managed care systems requiring specialists to (a) choose to be either wholly a specialist or wholly a primary care physician or (b) elect to do both. Finally, the last study of primary care delivered by specialists and subspecialists was the Mendenhall study in the 1970s (Mendenhall et al., 1978a, 1978b; Aiken et al., 1979), when physician supply, health care organization and financing, and other aspects of the U.S. health care system were quite different from those that obtain today. On the grounds of timeliness alone, revisiting these important questions can be justified.
Priority Areas Based on the IOM Definition
If the capacity to investigate primary care were to exist, what would it do? This was a major focus of the workshop organized by this committee in January 1995. To some extent, workshop results echoed earlier AHCPR findings (Mayfield and Grady, 1990; AHCPR Task Force, 1993). The remainder of this chapter identifies high-priority research topics tied to core elements of the IOM definition of primary care in Chapter 2.8
Large Majority of Health Care Needs
Documenting and getting consensus on what constitutes the "large majority of personal health care needs" in this country are formidable tasks. The questions (and the answers) differ by population and individual patient—for instance, families without children in contrast to those with children; patients with a chronic disease who do, or do not, have significant comorbid conditions; and rural, suburban, and inner city populations. Different ethnic groups, communities, and individuals may have quite different perceptions of illness and care-seeking behaviors, meaning that uniform definitions of this concept may be difficult to arrive at. Finally, within the full range of health care needs, what rightly belongs in the primary care ambit and what can properly be regarded as specialty care deserves explication.
Health and health services should be broadly defined. Prevention is a principal element, so risk assessment and health risk appraisal are important. Both physical and mental health must be considered. Patients whose basic needs are
palliative or comfort-related (e.g., those with terminal illness), together with family members and other significant persons in their lives, are also an important population, even if the assistance lies more in social services, sophisticated pain management, or respite care than in traditional primary care.
Accessible and Integrated Health Care Services
Access to care. The IOM is on record as supporting universal access to insurance and health care (IOM, 1993a, p. 7): "All or virtually all persons—whether employed or not, whether ill or well, whether old or young—must participate in a health benefits plan." This committee, in identifying accessibility as a core element of primary care, subscribes to this same goal (see Chapter 5). The nation is moving in the opposite direction, however; for instance, Short and Banthin (1995) report that, of persons under 65 years of age who are insured, more than 22 percent are now underinsured for all or part of a year; overall, more than 35 percent of the nonelderly (between 75 million and 79 million persons) have inadequate health insurance today.
Primary care research must continue to examine patterns of use of services by all members of a given population (e.g., a community, a managed care plan, or special groups such as the homeless), with particular emphasis on the care provided to the uninsured and the underserved. The national survey-based data set on primary care recommended earlier would help provide information on (a sample of) all individuals, not simply on users of services. Several special aspects of the access issue warrant early attention:
- To what extent are physical health and mental health concerns and services treated in a parallel manner? More specifically, are services for mental disorders and substance abuse given reasonable parity with those for physical ailments?
- To what degree can certain populations gain access to services in "nontraditional" settings? For example, how available in school-based settings are counseling or preventive services for sexually active teenagers?
Integration of health care services—coordination and continuity. Methodologically, questions in this area lend themselves best to analyses of episodes of care, and practice-based networks may have a comparative advantage for such work. Among the key issues:
- What are the most efficient and highest quality ways to coordinate and combine health care services, taking time, type of provider, type of setting, and patient disease and sociodemographic characteristics into account? For example, what works best for major chronic conditions (e.g., diabetes, multiple sclerosis)?
- elective or emergency surgical procedures (e.g., hip replacement)? mental or cognitive problems (e.g., dementia)?
- What impact on coordination and continuity does comorbidity have? For instance, how best might integration of services be accomplished for acute problems among patients with long-standing chronic illness? Would the answers differ for the major causes of morbidity and mortality in this country, such as cancer, cardiovascular disease, acquired immunodeficiency syndrome (AIDS), or violence-related problems among the young?
- What distinguishes an effective primary care team from an ineffective one? What impact does team delivery have on costs, quality, and access?
- What patterns of "substitution" of clinicians are emerging in managed care plans? in inpatient settings such as hospitals and nursing homes (IOM, 1996b)? What effect does such "substitution" have on patient outcomes and satisfaction with care? on provider and clinician satisfaction with performance?
A core element of primary care is the notion that individuals and families will have a "usual clinician" (i.e., physician, nurse practitioner, or physician assistant) with whom they will have a long-standing relationship based on mutual trust and respect. For today's health care market, however, several issues emerge:
- Do most people today have a usual clinician or source of personal health care? Does the answer to this still depend heavily on insurance status and, if so, how?
- What types of usual clinician do various people prefer? What types of clinician do they actually have?
- What responsibilities do patients expect such clinicians to have? What do clinicians themselves see as their responsibilities as a usual source of care?
Part of the appeal of a sustained partnership is the belief that clinicians will come to a deep understanding of their patients' preferences (or, in the research vernacular, utilities) for different health outcomes. Among the topics of interest:
- Does understanding patient preferences derive from, and contribute to, better communication between practitioners and patients? Does it need to involve communication between members of a primary care team? between team members (on the one hand) and families (on the other)?
- Does a better understanding of patient utilities make any difference in the types or intensity of services offered? in the kinds of services demanded or accepted by patients and families? in the health outcomes achieved? For instance, what is the effect of trust (manifested through sustained partnerships) in
- improving the decisionmaking and outcomes in "difficult" diseases or circumstances, such as schizophrenia or at the end of life?
Accountability for Quality of Care
Traditionally, among the most difficult and contentious areas to conceptualize and measure is quality of care. This becomes especially true as the managed care revolution begins to shift incentives away from the utilization- and cost-inducing ones of the country's traditional fee-for-service system of health care and toward the utilizationand cost-constraining incentives of managed care, and as the language turns more toward informed purchasing and accountability and away from quality improvement. These shifts pose numerous questions, including simply defining accountability, performance monitoring, informed purchasing and clarifying how they are similar to, and how different from, quality assessment and quality improvement.
Measuring and improving quality of care and evaluating quality assurance and quality management programs in health care are crucial to primary care. In 1990, an IOM committee concerned with the Medicare program's quality assurance strategy laid out a research agenda for quality of care that categorized research priorities into one of three stages: basic research, applied research, and diffusion. It called for
basic research [in] the following topics: (1) variations, effectiveness, and appropriateness of medical care interventions; (2) process-of-care measures for both the technical aspects of care and the art of care; (3) outcomes, health status, and quality of life; and (4) continuous improvement models. Priorities in applied research included: (1) linking process and outcomes; (2) practice guidelines; (3) effectiveness of quality assurance interventions; (4) various setting-specific issues (relating to hospitals, ambulatory care, home health care, and HMOs); (5) rural health care; and (6) the effects of organizational and financing arrangements on quality of care and quality assurance. Finally, with respect to diffusion, … the following areas [warrant] continued work and investigation: (1) data systems and hardware; (2) data sharing; (3) data feedback and disclosure; and (4) program evaluation [IOM, 1990, p. 364].
To this broad agenda can be added:
- How best should patient (consumer, purchaser) satisfaction be measured? clinician satisfaction?
- Do patient and clinician satisfaction interact and, if so, how?
- In today's rapidly changing market, who has final responsibility for decisions about adopting new technologies? for abandoning obsolete technologies?
- As more and more tasks are delegated across members of the primary care team, what are necessary skills for use of various technologies and procedures? How can appropriate levels and modes of training help to ensure those skills?
Finding the answers to ethical dilemmas is among the greatest challenges in health care today. It may prove especially demanding for those in the front lines—that is, primary care. Even posing the questions about the mission of primary care, research priorities, and criteria for quality assessment can be difficult when different value systems and models of health care delivery intersect (Lamberts and Hofmans-Okkes, 1996b). Among the questions are:
- Who exactly is accountable to whom in the health plan-clinician-patient triad?
- Are physicians or other clinicians now employed by managed care plans and integrated delivery systems finding themselves in a "dual agent" position, ostensibly accountable not just to their patients but also to the plans and systems within which they practice? How are they to reconcile their duties to systems and to individual patients? How should primary care clinicians harmonize either of those sets of obligations to those posed by society at large?
Family and Community Context
In this committee's view, primary care must be practiced in the context of family and community. In research terms, numerous issues arise:
- How should a "family" be defined or characterized today? a "community"?
- How can proper account be taken of cultural and ethnic differences in a society as diverse as that of the United States?
- What constitutes "dysfunctional" families and communities, and how might these factors affect the way primary care is practiced and the health outcomes expected?
- What role do families and external social support systems play in episodes of care for a particular individual? For instance, what is the impact of social support networks on family functioning and patient outcomes? What role should primary care clinicians play in mobilizing family (or outside) support and involvement in the ongoing care for individuals? What role do they play?
The committee gave special attention to the bond between primary care and public health (see Chapter 5). For research purposes, three sets of topics warrant focused attention:
- What, in today's fiscal and political environment, does the link between primary health care and public health functions actually mean in practical terms? Does this differ by geographic area? political jurisdiction?
- How permeable are the boundaries between health care and public health?
- What "clusters" of problems (e.g., infectious disease, violence, or trauma) can be identified within families? within communities?
Primary care in the United States represents a largely uncharted frontier, awaiting discovery and exploration. Expanded research in this area is timely because of the accelerating movement toward a variety of managed care and integrated delivery systems, most of which will rely increasingly on primary care models and clinicians. To the degree that this is so, improved primary care that can bring about a better balance between patients' and populations' needs and the health care services they receive is critical.
The science base for primary care is modest, and the infrastructure underlying the knowledge base is skeletal at best. Thus, the committee in this chapter has advanced four recommendations intended to strengthen the underpinnings of a primary care research enterprise. Those relate to (1) federal support for primary care research, including the designation of a lead agency in this effort; (2) development of a national database on primary care, ideally through some form of ongoing survey mechanism; (3) support of research through primary care practice-based research networks; and (4) development of standards for data collection, including attention to data element definition and improved coding. The committee also identified several subjects that it believes warrant high priority in any primary care research agenda. Prominent among these was the committee's recommended study of specialist provision of primary care. Other areas involve major elements of the committee's conceptualization of primary care, such as the large majority of personal health care needs, accountability, and practicing in a family and community context.
This chapter has pulled together the threads of the committee's views about the value, nature, and delivery of primary care into a formidable research agenda and a call for infrastructure development. The final chapter of this report takes those themes, together with the issues raised about the primary care workforce and education and training, to offer the committee's views about how its vision of the future of primary care might be implemented in coming years.
AHCPR Task Force. Putting Research into Practice. Report of the Task Force on Building Capacity for Research in Primary Care. AHCPR Publication No. 94–0062. Rockville, Md.: Agency for Health Care Policy and Research, Department of Health and Human Services, August 1993.
Aiken, L.H., Lewis, C.E., Craig, J., et al. The Contribution of Specialists to the Delivery of Primary Care. New England Journal of Medicine 300:1363–1370, 1979.
CBASSE/IOM (Commission on Behavioral and Social Sciences and Education and Institute of Medicine). Toward A National Health Care Survey: A Data System for the 21st Century. G.S. Wunderlich, ed. Washington, D.C.: National Academy Press, 1992.
Donaldson, M.S., and Vanselow, N.A. The Nature of Primary Care. Journal of Family Practice 42:113–116, 1996.
Federal Register. Department of Health and Human Services, Office of the Secretary. Agency for Health Care Policy and Research; General Reorganization; Statement of Organization, Functions, and Delegations of Authority. Federal Register 60:37898–37900, July 24, 1995.
Green, L.A. Science and the Future of Primary Care. Journal of Family Practice 42:119–122, 1996.
Hornbrook, M.C., Hurtado, R.V., and Johnson, R.E. Health Care Episodes. Definition, Measurement, and Use. Medical Care Review 42:163–218, 1985.
Inui, T.S. What Are the Sciences of Relationship-Centered Primary Care? Journal of Family Practice 42:171–177, 1996.
Journal of Family Practice 42:113–203, 1996.
IOM (Institute of Medicine). Medicare: A Strategy for Quality Assurance . K.N. Lohr, ed. Washington, D.C.: National Academy Press, 1990.
IOM. The Computer-Based Patient Record: An Essential Technology for Health Care. R. Dick and E. Steen, eds. Washington, D.C.: National Academy Press , 1991a.
IOM. Extending Life, Enhancing Life: A National Research Agenda on Aging. E.T. Lonergan, ed. Washington, DC: National Academy Press, 1991b.
IOM. Assessing Health Care Reform. M.J. Field, K.N. Lohr, and K.D. Yordy, eds. Washington, D.C.: National Academy Press, 1993a.
IOM. Emergency Medical Services for Children. J.S. Durch and K.N Lohr, eds. Washington, D.C.: National Academy Press, 1993b.
IOM. Defining Primary Care: An Interim Report. M. Donaldson, K. Yordy, and N. Vanselow, eds. Washington, D.C.:National Academy Press, 1994a.
IOM. Health Data in the Information Age: Use, Disclosure, and Privacy . M.S. Donaldson and K.N. Lohr, eds. Washington, D.C.: National Academy Press, 1994b.
IOM. Health Services Research: Work Force and Educational Issues. M.J. Field, R.E. Tranquada, and J.C. Feasley, eds. Washington, D.C.: National Academy Press, 1995.
IOM. The Nation's Physician Workforce: Options for Balancing Supply and Requirements. K.N. Lohr, N.A. Vanselow, and D.E. Detmer, eds. Washington, D.C.: National Academy Press, 1996a.
IOM. Nursing Staff in Hospitals and Nursing Homes: Is It Adequate? G.S. Wunderlich, F. Sloan, and C. Davis, eds. Washington, D.C.: National Academy Press, 1996b.
Keeler, E.B., Rolph, J.E., et al. The Demand for Episodes of Medical Treatment: Interim Results from the Health Insurance Experiment. R-2829-HHS. Santa Monica, Calif.: RAND Corporation, 1982.
Lamberts, H., and Hofmans-Okkes, I. Episode of Care: A Core Concept in Family Practice. Journal of Family Practice 42:161–167, 1996a.
Lamberts, H., and Hofmans-Okkes, I. Values and Roles in Primary Care. Journal of Family Practice 42:178–180, 1996b.
Lamberts, H., Wood, M., and Hofmans-Okkes, I., eds. The International Classification of Primary Care in the European Community with a Multi-Language Layer. New York: Oxford University Press, 1993.
Lohr, K.N., Brook, R.H., Kamberg, C.J., et al. Use of Medical Care in the Rand Health Insurance Experiment: Diagnosis- and Service-Specific Analyses in a Randomized Controlled Trial. Medical Care 24(9 Suppl.):S1–S87, 1986.
Mayfield, J., and Grady, M.L., eds. Primary Care Research: An Agenda for the 90s. DHHS Publication No. (PHS) 90–3460. Rockville, Md.: Agency for Health Care Policy and Research, September 1990.
Mendenhall, R.C., Girard, R.A., and Abrahamson, S. A National Study of Medical and Surgical Specialties. I. Background, Purpose, and Methodology. Journal of the American Medical Association 240:848–852, 1978a.
Mendenhall, R.C., Lloyd, J.S., Repicky, P.A., et al. A National Study of Medical and Surgical Specialties. II. Description of the Survey Instrument. Journal of the American Medical Association 240:1160–1168, 1978b.
Nerenz, D.R. Primary Care Research from a Health Systems Perspective. Journal of Family Practice 42:186–191, 1996.
Newhouse, J.P., and the Insurance Experiment Group. Free for All? Lessons from the RAND Health Insurance Experiment. Cambridge, Mass.: Harvard University Press, 1993.
Niebauer, L.J., and Nutting, P.A. Primary Care Practice-Based Research Networks Active in North America . Journal of Family Practice 38:425–426, 1994.
Nutting, P.A. Practice-Based Research Networks: Building the Infrastructure of Primary Care Research. Journal of Family Practice 42:199–203, 1996.
Perrin, E.B. SAC Instrument Review Process. Medical Outcomes Trust Bulletin 3(4):1, September 1995.
Povar, G.J. Primary Care: Questions Raised by a Definition. Journal of Family Practice 42:124–128, 1996.
Scientific Advisory Committee. Instrument Review Criteria. Medical Outcomes Trust Bulletin 3(4):I–IV, September 1995.
Short, P.F., and Banthin, J.S. Caring for the Uninsured and Underinsured: New Estimates of the Underinsured Younger Than 65 Years. Journal of the American Medical Association 274:1302–1306, 1995.
Stange, K.C. Primary Care Research: Barriers and Opportunities. Journal of Family Practice 42:192–198, 1996.
Starfield, B. A Framework for Primary Care Research. Journal of Family Practice 42:181–185, 1996.
Williams, K.N., and Brook, R.H. Research Opportunities in Primary Care. The Mount Sinai Journal of Medicine 45:663–672, 1978.