George A. Goldberg

Does free care improve adults' health? Results from a randomized controlled trial.

Does free medical care lead to better health than insurance plans that require the patient to shoulder part of the cost? In an effort to answer this question, we studied 3958 people between the ages of 14 and 61 who were free of disability that precluded work and had been randomly assigned to a set of insurance plans for three or five years. One plan provided free care; the others required enrollees to pay a share of their medical bills. As previously reported, patients in the latter group made approximately one-third fewer visits to a physician and were hospitalized about one-third less often. For persons with poor vision and for low-income persons with high blood pressure, free care brought an improvement (vision better by 0.2 Snellen lines, diastolic blood pressure lower by 3 mm Hg); better control of blood pressure reduced the calculated risk of early death among those at high risk. For the average participant, as well as for subgroups differing in income and initial health status, no significant effects were detected on eight other measures of health status and health habits. Confidence intervals for these eight measures were sufficiently narrow to rule out all but a minimal influence, favorable or adverse, of free care for the average participant. For some measures of health in subgroups of the population, however, the broader confidence intervals make this conclusion less certain.

COMPARISON OF HEALTH OUTCOMES AT A HEALTH MAINTENANCE ORGANISATION WITH THOSE OF FEE-FOR-SERVICE CARE

To determine whether health outcomes in a health maintenance organisation (HMO) differed from those in the fee-for-service (FFS) system, 1673 individuals ages 14 to 61 were randomly assigned to one HMO or to an FFS insurance plan in Seattle, Washington for 3 or 5 years. For non-poor individuals assigned to the HMO who were initially in good health there were no adverse effects. Health outcomes in the two systems of care differed for high and low income individuals who began the experiment with health problems. For the high income initially sick group, the HMO produced significant improvements in cholesterol levels and in general health ratings by comparison with free FFS care. The low income initially sick group assigned to the HMO reported significantly more bed-days per year due to poor health and more serious symptoms than those assigned free FFS care, and a greater risk of dying by comparison with pay FFS plans.

The effect of cost sharing on the use of antibiotics in ambulatory care: Results from a population-based randomized controlled trial,

Little is known about how generosity of insurance and population characteristics affect quantity or appropriateness of antibiotic use. Using insurance claims for antibiotics from 5765 non-elderly people who lived in six sites in the United States and were randomly assigned to insurance plans varying by level of cost-sharing, we describe how antibiotic use varies by insurance plan, diagnosis and health status, geographic area, and demographic characteristics. People with free medical care used 85% more antibiotics than those required to pay some portion of their medical bills (controlling for all other variables). Antibiotic use was significantly more common among women, the very young, patients with poorer health, and persons with higher income. Use of antibiotics for viral, viral-bacterial, and bacterial conditions did not differ between free and cost-sharing insurance plans, given antibiotics were the treatment of choice. Cost sharing reduced inappropriate and appropriate antibiotic use to a similar degree.

Management of patients on psychotropic drugs in primary care clinics.

While nonpsychiatrist physicians account for the majority of prescriptions written for psychotropic drugs, little is known about the quality of their drug management strategy. We studied this issue using data from 16 academic internal medicine group practices. Data on treatment, abstracted from medical records, were compared to criteria for quality care. Eighteen percent of patients used minor tranquilizers or antidepressants. The only individual factor independently associated with use of minor tranquilizers was mental health status. Nonwhites were less likely than whites to be diagnosed as depressed or receive antidepressants, even after controlling for baseline mental and physical health status. Mental and physical health status were also independently associated with antidepressant drug use. Quality of care was low for formulating a treatment plan for either drug group and for follow-up plans for antidepressants. Documentation of an adequate treatment plan for minor tranquilizers was poorest for patients who visited a house staff or nonphysician rather than a faculty member. For antidepressants, the patients with the poorest general health status tended to have the best documentation of treatment plans.

Quality of ambulatory care. Epidemiology and comparison by insurance status and income.

In this report the data from medical history questionnaires, screening examinations, insurance claims, and a face–to–face physician interview were used to examine the quality of ambulatory care received for 17 chronic conditions by a general population of 5986 adults (≤65) and children (≤14) enrolled in the RAND Health Insurance Experiment. Subjects in six U.S. sites were randomly assigned to insurance plans that were free or that required cost sharing, or in one site to an HMO. Quality–of–care criteria—both process (what was done to patients) and outcome (what happened to them)—were developed. Overall, 81% of outcome criteria and 62% of process criteria were met. Physicians interviewed patients with selected conditions at the Experiments end to evaluate care. They suggested that approximately 70% of patients should have their current therapy changed, but only 30% of patients would obtain more than minor improvement from such a change. Clinically meaningful plan differences in quality of care were observed only for the process criteria dealing with the need for a visit (free plan compliance 59%; cost sharing compliance 52%). Quality of care for the poor was slightly worse than for the nonpoor and persons randomized to an HMO had slightly better overall quality of care than those in the fee–for–service system. Substantial improvements in the quality of the process of care could be made, but impact on outcome may be small. Results of the analysis suggest the need for development of clinical models to test the relationship between specific process criteria and improvements in outcome.

Effect of a Health Maintenance Organization on Physiologic Health: Results from a Randomized Trial

In a previous comparison of persons between 14 and 62 years of age randomly assigned to receive care through a fee-for-service system (n = 784) or through a health maintenance organization (HMO) (n = 738) in Seattle, Washington, persons in the HMO had much lower hospital expenditures and admissions, more bed days, a higher prevalence of serious symptoms, and less satisfaction with care. We report an examination of 20 additional health status measures. Our results are consistent with a hypothesis of no differences in health status measures between the two systems. In addition, a comparison of nine health practices between the systems also indicated no overall differences. Most physiologic measures and health practices for a typical person were not affected by care received through the fee-for-service system or the HMO. However, we are less certain of this result in specific subgroups, such as persons of lower income initially at elevated risk, because confidence intervals are necessarily wider. We conclude that the cost savings achieved by this HMO through lower hospitalization rates were not reflected in lower levels of health status.

Comparison of a Criteria Map to a Criteria List in Quality-of-Care Assessment for Patients with Chest Pain: The Relation of Each to Outcome

In a prospective study we compared the ability of two quality assessment methods—the standard criteria list and the criteria map—to predict the appropriateness of the disposition decision for 421 patients with chest pain who presented to two emergency departments. To evaluate the quality of this decision, each patient was followed at home or in the hospital to determine whether an acute condition requiring hospital admission was present. Among the 169 discharged patients, the map scores of the eight with admissible disease were significantly higher than the score for those without admissible disease (p = 0.02). For the 252 admitted patients, a similar relationship between map score and the admissible disease outcome was observed (p = 0.0001). There was no significant relationship between list score and outcome among either the admitted or the discharged patients. Multivariate logistic analyses confirmed the importance of the map score as a predictor of admissible disease. The map score was superior to the list score and to demographic variables in its ability to correctly classify patients with and without admissible disease. The demonstrated relationship between map score and patient outcome enables the map to be used in a quality assurance system. An institution can ensure that physicians review an enriched sample of the inappropriate discharges and the unjustified admissions by selecting admitted patients with low map scores and discharged patients with high map scores.

Providing Primary General Medical Care in University Hospitals: Efficiency and Cost

Data on efficiency, costs, and profits of 15 internal medicine outpatient group practices in university hospitals were collected for 9 months from interviews, a time-motion study, observations, and reviews of bills. Charges for a follow-up visit were about 25% higher than Medicares allowable charges, but differed threefold across practices. Physicians spent more than half their allocated patient care or supervision time in other activities and 14% of nursing time was used for direct patient care. Visits to second- and third-year residents cost one half of those to faculty. Faculty supervision of second- and third-year residents was limited; it was, on average, 2 minutes per follow-up visit. Despite these inefficiencies, bad debts, and educational costs, practices appeared to break even financially. We conclude it is financially feasible for university hospitals to provide primary care to disadvantaged populations.

Explicit Criteria for Use of Laboratory Tests

Excerpt To the editor: Several studies have pointed out or implied overuse of laboratory tests by physicians (1-3). However, such overuse has not been defined objectively or demonstrated explicitly...

Development of hospital levels of care criteria.

A modification of the Nominal Group/Delphi Technique can be applied to the problem of developing among physicians levels-of-care criteria for use in “concurrent review” activities. UR requirements can be rationalized through a process that uses physicians’ scare time effectively, allows for cyclical feedback on their peers’ views, and insures input from both dominant and reticent individuals.