Benchmarking medical group practices using claims data: methodological and practical problems.

As claims data for physicians and groups of physicians has improved in quality and quantity, health information vendors have begun marketing information about medical groups' productivity, utilization, and quality. Based on interviews with product developers and our understanding of the evolution of their products, several methodological and practical issues remain. For now and the immediate future, health information vendors will continue to face the limitations of physicians' claims data. Vendors and purchasers should be aware of common data shortcomings such as inadequate monthly enrollment figures, possible physician upcoding to circumvent utilization management restrictions, and incorrect coding when a test is used to rule out a disease. In the longer term, several avenues seem likely to make medical groups' data better and richer because of computer-based medical records and efficiencies possible from the Internet. The field of benchmarking products for group practices is still an immature market. However, several trends suggest such products are highly desirable. Provider organizations which bear medical risk need benchmarking data to help improve their efficiency. There are many important nonprovider organizations that need good information on group practices' utilization patterns and outcomes to help them plan new products and negotiate with physicians.

Episode Treatment Groups (ETGs): a patient classification system for measuring outcomes performance by episode of illness.

The Episode Treatment Group (ETG) methodology is a patented case-mix adjustment and episode-building system that uses routinely collected inpatient and ambulatory claims data. The resulting clinically homogenous groups, of which there are approximately 600, adjust for severity by the presence of complicating conditions, comorbidities, and other characteristics of a patient's condition that affect resource utilization.

Helping financial analysts communicate variance analysis.

Healthcare organizations often use variance analysis to explain variation between planned and actual costs and charges. This type of analysis is becoming even more common as healthcare executives work to improve efficiency, to set priorities for organizational improvement as part of strategic planning, and to explain costs and charges to interested groups such as purchasers and payers. Variance analysis produces data that must be presented in a format useful to senior executives. An effective format would express the data in a visual summary that is simple enough to be readily understood and detailed enough to provide valuable information.

Use of the resource-based relative value scale for private insurers.

Medicare's resource-based relative value scale (RBRVS) was implemented 1 January 1992 for physician payment using a conversion factor of $31 for each relative value unit (RVU). We calculated a conversion factor of $42.24 for The Travelers Insurance Company's group health plan business using the RBRVS methodology and the calendar year 1990 Travelers Large Case Norms Extract of active employees. This DataWatch describes two important applications of the relative value scale for private insurers: for pricing and for analyzing claims expenditures.

Identifying patients who need social work services: an interdisciplinary analysis.

Factors that affect decisions made by professionals in the referral of hospital patients for social work services are analyzed. A multidisciplinary approach was used, in which social workers, nurses, and physicians individually evaluated the social service needs of a sample of patients hospitalized in an intermediate care section at a Veterans Administration medical center. The study helped to identify sources of divergent views held by health care professionals in different disciplines.

Evaluating and predicting outcome of acute cerebral vascular accident.

An innovative evaluation method is used to study the outcomes and clinical predictions for 97 patients with acute cerebral vascular accidents. The technique involved the participation of several professional disciplines in selecting baseline and treatment variables and making independent predictions about the functional status of patients upon discharge from the stroke treatment center. The data suggest that (1) baseline variables were more important than treatment variables in the participants' predictions about the patients' short-term outcomes; (2) stroke unit staff members were generally successful in predicting patients' functional status; and (3) stroke extensions and other complications are important factors which affect stroke patients' short-term outcomes.

High users of VA emergency room facilities: are outpatients abusing the system or is the system abusing them?

A phenomenon well known to emergency room personnel is the high use of ER facilities by a small number of patients. In this study of 335 patients followed in outpatient specialty clinics at a university-affiliated VA medical center, 23% of the patients accounted for 73% of the ER visits. Although some patients may be abusing the system, the problem is difficult to correct because of congressional legislation that deters the VA from providing primary care. Thus, a small subset of patients with chronic medical problems who live close to the hospital are likely to continue to consume a disproportionate amount of ER resources.

The usefulness of patients' individual characteristics in predicting no-shows in outpatient clinics.

Patients' characteristics and no-show patterns are analyzed in order to determine the number of patients to schedule in outpatient clinics. This predictive model is evaluated on a second sample of data and compared with another scheduling technique that is based on the average no-show rate for each clinic. Our survey suggests that it is possible to predict accurately the number of no-shows with a small set of variables, and that patient scheduling can be improved by paying attention to the characteristics of individual patients. The most important single predictor is the patient's previous appointment-keeping pattern.

Duration of blood pressure elevation in accurately predicting surgical cure of renovascular hypertension.

Among 110 patients who underwent corrective surgery for unilateral renovascular hypertension, we found that a preoperative of hypertension was a highly important predictor of postoperative achievement of normotension. Those with less than a 5-year history of hypertension experienced 78% incidence of successful outcome, compared to such a salutary frequency of only 25% in patients with longer hypertension duration. Although the best renal vein renin (RVR) boundary ratio (1.4) was less predictive of overall surgical success in the population studied, the prognostic value of this test improved considerably when analysis of RVR ratio results were confined to patients not receiving renin-suppressing agents during RVR sampling and who had technically satisfactory operations. Highest surgical benefit rate occurred in the group of patient with both short duration of hypertension and high RVR ratio. Conversely, patients with long hypertension duration and low RVR ratio exhibited lowest surgical success frequency. Therefore, duration of hypertension is hereby shown to be an important factor in the preoperative evaluation of appropriate management of patients with renovascular hypertension.

The trips of slips. Time delays in laboratory and X-ray data for outpatients in a teaching hospital.

The laboratory and x-ray studies performed on 148 outpatients were monitored to see how long it took for the results to be obtained, reviewed, and inserted into the medical record. An average of 27.0 days was required for all of these tasks. This mundane, unwieldly problem clearly deserves more attention, especially in outpatient clinics at teaching hospitals.

A technique for analyzing clinical data to provide patient management guidelines.

This article describes a technique for analyzing clinical data in order to guide patient management decisions. The technique is illustrated by considering a specific decision problem encountered in the management of possible meningitis, namely, whether or not to administer antibiotics while awaiting the results of a CSF bacterial culture. Data from 303 patients with meningitis are analyzed in order to determine which combination of clinical variables best discriminates between bacterial and aseptic cases. From these variables, a probability tree is constructed that indicates the probability of bacterial meningitis, depending on a patient's clinical characteristics. In addition to identifying the most important variables, the analysis reveals that a number of tests performed routinely on patients with meningitis are of questionable diagnostic value.