Correspondence
A Trial of Increased Access to Primary Care
N Engl J Med 1996; 335:895-898September 19, 1996
- Article
To the Editor:
The elderly patients studied by Weinberger and colleagues (May 30 issue)1 may have had additional access to physicians and nurses during the six-month period after their discharge from the hospital, but the question remains: Did they receive primary care? Too often, what is called primary care lacks the critical component of continuity of care by a physician who is specifically trained and skilled in the comprehensive first contact with and ongoing care of a population.2 Table 1 of Weinberger et al. shows that members of the study group were assigned to “primary care physicians,” whom they did not meet until shortly before hospital discharge. This mode of care might better be termed “discontinuity of care.” Physicians who practice primary care would have been caring for their patients throughout the hospital stay and probably for many years before. The mutual knowledge and experience of patient and physician in a continuous relationship is critical.
The study failed to define the primary care training or experience of the participating physicians, only three of whom were family physicians. How many were general internists and how many were actually subspecialty internists? Primary care physicians tend to moderate costs, whereas subspecialists raise them.3 The lack of a generally accepted definition of primary care keeps alive the notion that any physician or team of providers who cares for patients practices primary care.
The care provided in this study was misnamed; it was not primary care but rather early intervention in the treatment of discharged patients.
3 ReferencesDouglas E. Henley, M.D.
American Academy of Family Physicians, Fayetteville, NC 283051
Weinberger M ,Oddone EZ ,Henderson WG . Does increased access to primary care reduce hospital readmissions? N Engl J Med 1996;334:1441-1447
Full Text | Web of Science | Medline2
Kahn NB Jr ,Ostergaard DJ ,Graham R . AAFP constructs definitions related to primary care. Am Fam Physician 1994;50:1211, 1214-1215, 1218
Web of Science3
Moore GT . The case of the disappearing generalist: does it need to be solved? Milbank Q 1992;70:361-379
CrossRef | Web of Science | Medline
To the Editor:
Weinberger et al. show that intensified ambulatory services (characterized in their paper as “primary care”) do not reduce hospitalization rates for sick patients. This article and the accompanying editorial will receive great attention at a time when the stated objective of the Department of Veterans Affairs is to provide comprehensive primary care for all veterans.1 However, the intervention cannot be characterized as “primary care,” which entails comprehensive, continuous, longitudinal services to sick and well populations. In fact, people with existing sources of primary care were explicitly excluded from the study. Assigning a patient to a primary care physician does not achieve primary care, which requires a relationship developed over time between a provider and a patient. It takes much longer than six months to achieve the benefits of this “longitudinality.”2
Providing patients with increased access to clinical services after they have become seriously ill occurs in two ways: by design or by necessity. In either case, patients will find their way into the hospital. Case management for patients as sick as the ones in this study requires far more extensive interventions than the one used.
At its most basic level, “primary care” is a strategy for attending to all common needs of populations through the development of long-term person-focused (not disease-focused) relationships between providers and patients.3 Once patients become seriously ill, services must go beyond primary care. Here, effective case management becomes essential. The findings of Weinberger et al. suggest the need for better case-management strategies in the Department of Veterans Affairs, including better alternatives to short-term hospitalization. Their study does not address primary care.
3 ReferencesBarbara Starfield, M.D., Ph.D.
Johns Hopkins University, Baltimore, MD 21205Thomas A. Parrino, M.D.
Brown University, Providence, RI 029121
Kizer K. Prescription for change. Washington, D.C.: Department of Veterans Affairs, 1996.
2
Hjortdahl P . Continuity of care: general practitioners' knowledge about, and sense of responsibility toward their patients. Fam Pract 1992;9:3-8
CrossRef | Web of Science | Medline3
Starfield B, Parrino TA, Headley F, Ashton C. Primary care in VA: a primer. Boston: Department of Veterans Affairs Management Decision Research Center, 1995.
To the Editor:
As interested specialists caring for older men in a large Veterans Affairs medical center, we think the authors might consider another possible explanation. A large proportion of ambulatory care in teaching hospitals, such as the nine participating here, has been traditionally provided by specialists. Could it be that in this study a shift in the balance between specialist and primary care led to the higher admission rate? Were the patients' connections to their specialists broken?
Although the authors showed that extra primary care worsened the ability of patients with severe illnesses to stay out of the hospital, we do not know whether increased involvement by specialists would have been more successful. As the authors noted, a specialist team can prevent readmissions in severely ill patients with congestive heart failure.1 The best and most cost-effective care for patients with severe chronic disease may well result when the knowledge and experience of both primary care providers and specialists are brought to bear.
1 ReferencesGordon L. Snider, M.D.
Wilson S. Colucci, M.D.
Clark T. Sawin, M.D.
Boston Veterans Affairs Medical Center, Boston, MA 021301
Rich MW ,Beckham V ,Wittenberg C ,Leven CL ,Freedland KE ,Carney RM . A multidisciplinary intervention to prevent the readmission of elderly patients with congestive heart failure. N Engl J Med 1995;333:1190-1195
Full Text | Web of Science | Medline
To the Editor:
Despite its title, the study by Weinberger et al. is almost irrelevant to primary care because the patient population was unrepresentative of typical primary care practice. It consisted of patients hospitalized in Veterans Affairs hospitals for end-stage chronic disease. “Patients were excluded if they were already receiving continuous care at a primary care clinic.” Of the 3209 patients eligible for randomization, 30 percent “declined to participate typical-ly . . . because they had an established relationship with a . . . physician.” Thus, the patients passed through a series of selection steps, whose net effect was to include only patients with severe chronic illnesses who had been least able to establish a relationship with a primary care provider. This study is not a test of the effectiveness of primary care. Nor is it in any sense a test of the effectiveness of “early intervention,” as suggested by the accompanying editorial.1
1 ReferencesMichael K. Magill, M.D.
Marc Babitz, M.D.
University of Utah School of Medicine, Salt Lake City, UT 84132Michael P. Silver, M.P.H.
HealthInsight, Salt Lake City, UT 841061
Welch HG . Questions about the value of early intervention. N Engl J Med 1996;334:1472-1473
Full Text | Web of Science | Medline
To the Editor:
The design of the study by Weinberger et al. appears to have resulted in a disruption in the continuity of care of the patients randomly assigned to the study intervention. Patients in this group were assigned to “new” primary care physicians, whereas patients in the control group were presumably followed by their regular care givers. Although most of the newly assigned primary care physicians visited the patients in the hospital shortly before discharge, this is still no substitute for a long-term physician–patient relationship. This lack of perspective on the long-term clinical course of the patients could easily have resulted in a lower threshold for rehospitalization, particularly during short-term followup. . . . It is also possible that the physicians who agreed to participate in the study were, by nature, more meticulous with respect to patient care and more cautious in deciding whether to readmit a given patient to the hospital. Other factors that could influence a physician's admission rate include age, sex, the number of years in practice, the number of years of postgraduate training, specialty or subspecialty, and board certification. It is impossible to know whether the reported differences in readmission rates were due to a failure of the intervention or merely to a difference in physicians' practice patterns.
Michael W. Rich, M.D.
Washington University School of Medicine, St. Louis, MO 63110To the Editor:
. . . Motivated by both the Hawthorne effect and the usual good intentions, the providers in the intervention reported by Weinberger et al. may have felt compelled to “do something.” Confronted by such seriously ill patients whose treatment histories had been selected for prior hospitalizations, one likely response was to hospitalize more. It is debatable whether this study measured anything more than an unbalanced action of the Hawthorne effect on providers in the intervention group.
George A. Corey, M.D.
University of Vermont, Burlington, VT 05405To the Editor:
More frequent primary care contacts are likely to result in more admissions. Imagine that a physician admits a patient to the hospital if the patient's health status falls below some critical threshold. Imagine that in chronic disease, the health status fluctuates, occasionally dipping below the physician's threshold. The likelihood of being hospitalized in a given period depends on two variables: the proportion of time the patient spends below the threshold, in turn a function of the patient's average health status, and the frequency of assessments. . . .
The findings of Weinberger et al. demonstrate neither that the patients in the intervention group received worse care, nor even that they did not receive better care. This study lends no support to Welch's (undeniably correct) claim in the accompanying editorial that “the doctrine of early intervention can sometimes be wrong.” It is an empirically testable question whether any intervention causes harm rather than benefit to patients. A separate question ought to be, If there is a benefit, at what cost? But defining benefits solely in terms of resources expended is circular and stacks the deck against giving more care. Hospitalization rates are a measure of the use of resources and a poor proxy for health status. The conclusion that more primary care leads to more resource utilization, although perhaps disappointing, says nothing about health status; the assumption that more resource utilization means worse health or worse care is dangerous. The only doctrine directly challenged by these findings is the naive belief that effective treatment and prevention will always save money.
Jeffrey H. Burack, M.D.
University of California, San Francisco, San Francisco, CA 94143-0320To the Editor:
. . . Welch in his editorial suggests that the intervention was potentially harmful, since the detection of health problems leads to more interventions, causing a “cascade effect” that could be harmful and without benefit to patients. However, the patients in the intervention group expressed a significantly higher level of satisfaction with medical care than the patients in the control group. The editorial claims that the interventions made the patients sicker, although there is no factual support for this contention. There was no difference in the quality of life and no statistical difference in mortality rates between the intervention and control groups.
The editorialist draws analogies between the results of the study and those of other studies on the outcome of early screening for breast and prostate cancer. Those studies are irrelevant to this one, which includes elderly patients with severe and serious illnesses, such as advanced-stage congestive heart failure, chronic obstructive pulmonary disease, and diabetes mellitus.
The study suggests that better care may not be cheaper but that it is more valuable, at least in the eyes of the patients. And the patients, not the health care planners or health insurers, are ultimately the best judges of the quality of care.1
1 ReferencesValery A. Portnoi, M.D.
Genesis Physician Services, Washington, DC 200371
LaCombe MA . What is it patients want? Am J Med 1995;99:588-589
CrossRef | Web of Science | Medline
To the Editor:
As the population continues to age, the proportion of our patients who are chronically ill will increase. Clarifying what is of value in caring for these patients is important. Often, the course of chronic illness can be modified only slightly by medical interventions. We may simply be unable to have a major impact on the rate of hospitalizations or mortality in many of these cases. A failure to show improvement in these outcomes does not necessarily imply that improved, more intensive primary care is not beneficial. When disease is chronic and not curable, the process of care assumes increased importance to patients as well as physicians — perhaps even greater importance than the outcomes of care.
In the study by Weinberger et al., there was really only one measure of the quality of the process of care: patient satisfaction. The satisfaction of patients with their medical care was markedly and significantly improved by the intervention. The importance of this should not be undervalued by those who wish to improve the care of the chronically ill.
Wendy L. Adams, M.D., M.P.H.
University of Nebraska, Omaha, NE 68198-5620Author/Editor Response
The authors reply:
To the Editor: Drs. Henley, Starfield and Parrino, and Magill et al. question whether the intervention was primary care. Certainly, we recognize that the patients in the study were atypical of those in many primary care practices: an established relationship with a physician did not exist before enrollment, and patients were veterans with severe chronic diseases. However, we disagree that the intervention was irrelevant to primary care. We examined the effect of initiating a program designed to provide continuous, coordinated, first-contact, longitudinal care to medically and socioeconomically vulnerable patients who lacked regular physicians. The clinicians, who overwhelmingly were board-certified general internists, had the requisite training, experience, and skills to deliver primary care to this patient population. To suggest that a preexisting established relationship was necessary would systematically exclude the provision of primary care to most vulnerable patients. We certainly recognize that six months may have been insufficient to reduce the use of health services. However, we agree wholeheartedly with Dr. Portnoi and Dr. Adams, who emphasize that the intervention resulted in a substantial improvement in patient satisfaction, an increasingly important marker of the quality of care for patients with late-stage chronic disease. Moreover, we agree with Dr. Burack, Dr. Portnoi, and Dr. Adams that questions about the effectiveness (and potential harm) of early intervention cannot be addressed by our study.
What then may explain our somewhat surprising findings? First, Snider et al. discuss the roles of generalists and specialists. Notably, patients in both groups had nearly identical use of subspecialists. Second, we believe it highly unlikely that long-standing patient–physician relationships were disrupted. Patients with such established relationships typically refused to participate. Third, Dr. Rich and Dr. Corey suggest that a Hawthorne effect may have been operating. Study physicians were not blinded to either the hypothesis or the fact that patients were in the intervention group. Thus, we were concerned that physicians might delay hospitalizations to support our hypothesis. However, it is unlikely that physicians increased admissions as a result of being observed. Finally, Dr. Burack offers an interesting hypothesis: to the extent that the health status of severely ill patients hovers around some threshold for hospital admission, increasing the contact between patients in the intervention group and their providers may have increased the probability of detecting the crossing of that threshold by any individual patient, thus resulting in increased readmissions.
Our results should generate future study of the structure of health care, particularly for medically and socioeconomically vulnerable patients. Snider et al. suggest a shared-care approach, in which generalists and specialists contribute to patient care. Other studies may examine the effect of external controls, such as preadmission approval. Well-designed studies, rather than untested assumptions, should be used to develop policy about the structure of health care.
Morris Weinberger, Ph.D.
Richard L. Roudebush Veterans Affairs Medical Center, Indianapolis, IN 46202Eugene Z. Oddone, M.D., M.H.Sc.
Durham Veterans Affairs Medical Center, Durham, NC 27705William G. Henderson, Ph.D.
Hines Veterans Affairs Medical Center, Hines, IL 60141Author/Editor Response
Drs. Starfield and Parrino and Magill et al. are correct when they say that the study by Weinberger et al. was not a trial of primary care as compared with something else. Some seem to have misperceived the title of the report, which made clear that it was about “increased access.” The study is best thought of as a trial of the intensity with which primary care is delivered.
A number of people have shared the concern of Magill et al. and Dr. Portnoi — How could a trial with such sick patients ever be construed as an “early intervention”? We usually associate that term with the screening of asymptomatic people, and it was in this context that the generic problems of early intervention were first recognized (e.g., more false positive results, pseudodisease, and lead-time and length biases). But patients do not have to be “healthy” for these problems to be relevant. They may come into play with any effort to advance the time of diagnosis.
How can one advance the time of diagnosis in a group of patients with chronic diabetes, congestive heart failure, and chronic obstructive pulmonary disease? Clearly, other diagnoses would have to be involved. Early and frequent follow-up in these patients could have advanced the time of diagnosis (and engendered early intervention) as the result of “classic screening” (i.e., for prostate cancer). Early diagnosis and intervention may have occurred with diagnoses related to the original diseases: perhaps skin infections or worsening nephropathy in the patients with diabetes, arrhythmia or worsening pulmonary edema in those with congestive heart failure, and pneumonia or lung cancer in those with chronic obstructive pulmonary disease. Although these patients were sick, there is plenty of room to find new diagnoses. Some may ultimately be important to the patients, but others may simply represent the health-status fluctuations described by Dr. Burack. Nevertheless, the group that had more frequent contact with their physicians might have had these diagnoses made earlier with earlier intervention, which was the rationale for the trial.
Dr. Burack suggests that hospitalization is only a measure of resource use. Surely it is more than that. Physicians generally admit patients because they believe patients are ill and will benefit from inpatient care. They would be remiss if they did otherwise, given the risks associated with hospitalization.1,2 Regardless of expenditures, increased hospitalization is not a neutral outcome.
2 ReferencesH. Gilbert Welch, M.D., M.P.H.
Department of Veterans Affairs, White River Junction, VT 050091
Creditor MC . Hazards of hospitalization of the elderly. Ann Intern Med 1993;118:219-223
Web of Science | Medline2
Brennan TA ,Leape LL ,Laird NM , et al. Incidence of adverse events and negligence in hospitalized patients -- results of the Harvard Medical Practice Study I. N Engl J Med 1991;324:370-376
Full Text | Web of Science | Medline
- Citing Articles (3)
Citing Articles
1
Carl van Walraven, Muhammad Mamdani, Jiming Fang, Peter C. Austin. (2004) Continuity of Care and Patient Outcomes After Hospital Discharge. Journal of General Internal Medicine 19:6, 624-631
CrossRef2
Susan A. Flocke, Kurt C. Stange, Stephen J. Zyzanski. (1998) The Association of Attributes of Primary Care With the Delivery of Clinical Preventive Services. Medical Care 36:SUPPLEMENT, AS21-AS30
CrossRef3
Steven Schreiber, Teresa Zielinski. (1997) The Meaning of Ambulatory Care Sensitive Admissions: Urban and Rural Perspectives. The Journal of Rural Health 13:4, 276-276
CrossRef
