Hellinger, F.J. (1996). "The impact of financial incentives on physician
behavior in
managed care plans: A review of the evidence," Medical Care Research and Review 53(3),
pp. 294-314.
This literature review examines the relationship between the financial incentives confronting
physicians in managed care plans and the utilization of services. It was conducted by Fred J.
Hellinger, Ph.D., of the Agency for Health Care Policy and Research's Center for Organization
and Delivery Studies. Dr. Hellinger's primary conclusion from the review is that the financial
incentives confronting physicians are a key element in explaining the lower utilization rates of
enrollees in managed care plans. However, this conclusion is not definitive because it is based on
studies that are subject to numerous sources of potential bias (e.g., patient selection effects,
physician selection effects, and missing variables). To isolate the impact of financial incentives
facing physicians on the performance of health plans, Dr. Hellinger points out that it will be
necessary to adjust for patient, physician, health plan, and market characteristics related to
utilization. Most studies conducted to date include little information about these characteristics,
and few health plans collect data on enrollee health status and attitudes toward health care.
Almost no health plans possess data on the health coverage of spouses or possess data on health
services consumed by enrollees that are covered by other insurers or paid out-of-pocket. In
addition, information about how each health plan reimburses a physician and the number of
patients in each health plan seen by a physician is needed to adjust for possible spillover effects
(i.e., financial incentives from one health plan may affect how a physician treats patients in other
health plans). The only means of obtaining this information is to contact each physician in the
study directly, but in almost all instances, this option is infeasible because it is too time consuming
and costly. Even if there was an attempt to obtain this information, it is unclear how many
physicians would respond to such a request or how accurate those who responded would be,
according to Dr. Hellinger.
Reprints (AHCPR Publication No. 97-R028) are available from the
AHCPR Publications Clearinghouse.
Midgette, A.S., Griffith, J.L., Califf, R.M., and others. (1996). "Prediction
of the
infarct-related artery in acute myocardial infarction by a scoring system using summary
ST-segment and T-wave changes." (AHCPR grant HS06208). American Journal of
Cardiology 78, pp. 389-395.
The left anterior descending (LAD) coronary artery, which provides 50 percent of myocardial
blood flow, is a critical artery for survival after acute myocardial infarction (AMI). And,
infarct-related artery location is the most important determinant of coronary ventricular function
in the early hours of an AMI. These researchers developed a scoring system to predict the artery
responsible for an AMI using ST-segment and T-wave changes on the initial electrocardiogram
(ECG). The system was based on data from 228 patients with AMI symptoms and tested in a
similar group of 223 patients. Using this system, the researchers accurately predicted the LAD,
right circumflex (RC), or left circumflex (LC) coronary artery as the infarct-related artery. In the
test patients, the system sensitivity and specificity, respectively, were 97 percent and 95 percent
for LAD lesions, 85 percent and 86 percent for RC lesions, and 73 percent and 60 percent for LC
lesions. The researchers conclude that information easily obtained on the ECG can accurately
predict the likelihood of the LAD, RC, or LC artery as the infarct-related artery. This may be
useful in the decision to administer thrombolytic (clot-busting) therapy.
Palmer, R.H., and Hargraves, J.L. (1996). "The ambulatory care medical
audit
demonstration project." (AHCPR grants HS03087 and HS05609). Medical Care 34(9),
pp.
SS12-SS28.
This paper describes the research design employed in the Ambulatory Care Medical Audit
Demonstration Project, a randomized, controlled trial of the impact of quality improvement
interventions on primary medical care in 16 group practices in Boston, which was conducted
between 1978 and 1983. The authors used a concurrent crossover design using randomized cycles
of quality assurance so that a practice was a control site for one clinical practice guideline and an
experimental site for another. For 12 months before and 18 months during and after quality
assurance experimental interventions, the authors measured practitioner conformance to
patient-care guidelines believed to improve outcomes. These included internal medicine
patient-care guidelines, for instance, followup of low hematocrit, cancer screening for women,
followup of high serum glucose, and monitoring of patients treated with digoxin, as well as four
pediatric patient-care guidelines, for example, followup of positive urine culture, screening for
disease and immunizing infants, management of acute gastroenteritis, and management of acute
ear infection.
Seneff, M.G., Zimmerman, J.E., Knaus, W.A., and others. (1996).
"Predicting the duration
of mechanical ventilation: The importance of disease and patient characteristics." (AHCPR
grant HS07137). Chest 110(2), pp. 469-479.
Intensive care unit (ICU) patients who require mechanical ventilation typically have a high
mortality rate and consume a disproportionate amount of nursing and financial resources. This
preliminary study demonstrates how an equation that predicts the duration of mechanical
ventilation, based on case-mix and a national database, can be used to compare ventilator
practices among ICUs with different case-mixes. Using the acute physiology and chronic health
evaluation (APACHE) III database, the researchers analyzed which variables were significantly
associated with the duration of mechanical ventilation. They calculated interhospital variations for
average durations of ventilation based on 5,915 patients in 42 ICUs at 40 U.S. hospitals. For
patients admitted to the ICU and ventilated on day 1, total duration of ventilation was primarily
determined by admitting diagnosis and degree of physiologic derangement as measured by the
acute physiology score. Average duration of ventilation for the 42 ICUs ranged from 2.6 to 7.9
days, but 60 percent of this variation was accounted for by differences in patient characteristics.
This equation will be useful for comparing ventilator practices between ICUs, as a quality
improvement mechanism, and for controlling for patient differences in clinical trials of new
therapies.
Shekelle, P.G., and Schriger, D.L. (1996, October). "Evaluating the use of
the
appropriateness method in the Agency for Health Care Policy and Research clinical
practice guideline development process." HSR: Health Services Research 31(4), pp.
453-468.
These members of the Acute Low Back Problems Guideline Panel, which was supported by the
Agency for Health Care Policy and Research, assessed the feasibility of the appropriateness
method in the AHCPR clinical practice guideline development process. They compared the results
of the appropriateness method with those obtained using evidence tables and informal consensus.
The appropriateness method elicits ratings of the "appropriateness" of performing medical
procedures from a group of expert clinicians for a comprehensive array of potential patient
indications and does not force consensus. The researchers created practice guideline statements
for four back pain topics using the conventional AHCPR method. Six months later, they created
new guideline statements for the topics using the appropriateness method. Results of the
appropriateness method for TENS, discography, and traction showed no disagreement among
panel members and no appropriate indications for their use in the patient scenarios considered.
These results are similar to the guideline statements produced using evidence tables and informal
consensus. However, clinical practice guideline statements about electrodiagnostics created from
appropriateness ratings were much more clinically specific than those created using evidence
tables and informal consensus. Neither informal consensus building nor the appropriateness
method was clearly preferred by a majority of panelists.
Zhou, X-H, Katz, B.P., Holleman, E., Melfi, C.A., and Dittus, R. "An
empirical Bayes
method for studying variation in knee replacement rates." (AHCPR PORT grant
HS06432). Statistics in Medicine 15, pp. 1875-1884.
Knee replacement surgery is an effective and commonly used surgical procedure to relieve the
pain and suffering associated with severe knee arthritis. However, large variations in the use of
this procedure, with no apparent geographic differences in disease prevalence or severity, may
represent a combination of under- and overutilization, both of which can have important effects
on the quality and costs of care. The authors used data from Medicare patients to develop
explanations for the variations and employed a two-stage approach to data analysis to account for
missing patient information. In the first stage, they used an extra Poisson regression to model
within-region variation of knee replacement rates, while adjusting for the type of patient
demographic information they had. In the second stage, they used an empirical Bayes method to
model between-region variation of knee replacement rates. This two-stage method is easily
implemented using the existing software package, GLIM. It may be useful for other studies when
the number of patients is so large that individual-level data would make the size of the data set
unmanageable.
Return to Contents
AHCPR Publication No. 97-0016
Current as of February 1997
Internet Citation:
Research Activities newsletter. January 1997, No. 200. AHCPR Publication No. 97-0016. Agency for Health Care Policy and Research, Rockville, MD. http://www.ahrq.gov/research/jan97/