Measuring the quality of ambulatory care.

Measuring the Quality of Ambulatory Care 

A patient seeks outpatient care for numerous reasons, generally entering the health care delivery system through a primary care provider. There are six major categories of encounters i this setting. The patient seeks medical attention for wellness care, periodic screenings, acute illnesses, follow up of acute illnesses, chronic illness surveillance, or psychological problems. While assessment of doctors often emphasizes their ability to treat acute illnesses, it is imperative that assessment be made of the physician's lerformance in all these encounters. In keeping with the academic mainstream of quality assurance review, they could be studied for structure, process, and outcome if possible. In an ambulatory care center, structure would include analysis of the office physical plant, ancillary personnel, and services. Process means the physician's diagnosis and therapeutic management. Outcome would measure the efficiency and effectiveness of structure and process. The matrix for this quality model is shown in figure 1, page 19.

There are a number of approaches presently being used to measure and evaluate physician performance, including patient satisfaction, utilization-efficiency data, medical management criteria achievement, patient severity assessment, and sentinel events analysis. The patient satisfaction approach assumes consumer wisdom and suggests that it correlates with the quality of medical care delivery. While this correlation is not linear, there is good evidence that patient satisfaction is a good tool in my quality assurance program. With the realization that over- or under-utilization of medical care is suboptimal, it is fair to use utilization-efficiency data as a fair market of physician performance.

Medical science has been struggling with medical management assessment. does the doctor make the right medical decision for each and every patient and malady? At present, evaluation of all these separate decisions is impossible. In fact, it may even be impossible to evaluate a statistically sufficient number of decisions to adequately measure this critical category. The best attempts i this category have been accomplished through establishing medical management audit criteria and measuring their achievement percentage. For example, if it is decided that patients with pharyngitis should receive an analysis for streptococcal disease, we can measure the percentage of time that this audit criteria is met.

It would be unfair to compare one physician to another without having a tool to compare their patient populations. It is necessary to develop an ambulatory care severity measure. At the present time, we may need to be satisfied with age and gender analysis, but the future may bring exquisite computerized methods of patient mix measurement utilizing encounter forms. Horn, whose efforts yielded an inpatient severity index (CSI), is working on a similar effort on the ambulatory side (ASI).[1]

Last, we can measure physician performance by in-depth peer review of generically screened sentinel events. The classic example is mortality conferences, which retrospectively evaluate the care delivered to a patient who dies. In this situation, the sentinel event is a death. But applying a system similar to Craddick's inpatient care analysis,[2] the Kaiser System has recently identified 23 sentinel events to be evaluated in its ambulatory programs in Hawaii and north Carolina.[3] It includes abnormal test results not addressed, drug reactions, complications of outpatient procedures, and patient transfers.

It is apparent in the matrix in figure 1 that certain measurement tools mentioned above are better than others for a particular square. For example, square 1 requires measurement of the quality of the structure of wellness examinations in the ambulatory office setting. Parameters such as patient satisfaction and medical management criteria achievement would be better suited for this then sentinel events analysis of utilization-efficiency data. In this category, we want to know if the doctor's office structure promotes wellness. Is is easy to see the doctor for well check-ups? Are efforts made by the doctor to remind or recall the patient when a check-up is due? When studying the outcome of chronically ill patients (square 15), sentinel events may be the best source of measurement. In this category, you could review th eambulatory records of hospitalized patients or patients who die to decide whether better ambulatory care could have been delivered to prevent these outcomes.

It is necessary to look at each of these squares and ask the question, "What do you want to know?" Then the major tools for evaluation and measurement need to be examined and selections made on the basis of resources and database. Only after we have a composite picture of all these matrix components will we have a true measurement of an ambulatory care center's ability to delivery high-quality health care. By measuring all these parameters, we will be able to pick out strengths and weaknesses of particular primary care physician offices and improve performance.

The matrix in figure 1 can be condensed. The categories of wellness and screening and of acute illness and follow up can be combined, reducing the matrix to 12 squares (figure 2, right). Each square is explored individually and mechanisms of evaluation are formulated in the bos on page 20. The final result for each ambulatory care center and/or physician would be a report card listing the percentageof criteria met successfully or a percentile ranking in each square and a total program assessment analysis (figure 3 right). The physician analyzed did well overall but out to look more closely at how physchological problems are handled.

Quality Measurement

Mechanisms

Within each matrix square, particular parameters need to be measured. To customize a program for a particular physician or facility, it may be necessary to use different surveillance techniques.

Office/Facility Site Visit--Assessments can be made through staff interviews, physician interviews, scheduling analysis, and physical plant evaluation. Many of these topics are covered during the credentialing process used by managed care plans.

Medical Audit Criteria--Minimum acceptable audit criteria can be used to evaluate the medical care process. We use a melding of standards suggested by medical societies and adjusted by regional peer review analysis.

Patient Survey--Remarkably in-depth information can be obtained from the consumer. nearly every square in the matrix can benefit from consumer satisfaction input. Patient surveys can have questions easily coded for input into each of these squares.

Utilization Statistics--All ambulatory care settings need tracking of utilization of specialists and hospital care, which can be followed by referral forms and/or hospital billing. Most industry analysis involves categories such as hospital days per year per 1000 patients or expenses incurred by specialty care per member per month or year.

Patient Complaint Log--Every medical facility should track patient concerns, suggestions, and complaints. Trend analysis may lead to quality improvements. Input from this tracking can be built in the matrix quantitative analysis.

Transfer Rates--All medical care sites should monitor the rate of transfer of patients, with emphasis placed on local changes. This is a quality measure, and exit surveys may be able to determine information that can be placed in the matrix analysis.

Generic Screens of Ambulatory Care Sentinel Events--These data are instrumental in evaluation of medical care process. With limited resources for focused chart review, this is an area where one can get a good return on time invested.

Exceptional Drug Activity Reports -- Many prescription drug programs now track medication usage by membership. When exceptional criteria are met, health plans and/or doctors are notified. This can be a particularly useful quality measurement parameter for the medical care process of chronically ill patients and psychosocial patient care.

Mental Health Provider Reports -- Mental health providers need to give feedback to referring physicians. At the very least, we need to know whether the referrals were appropriate. Mechanisms for this analysis need more development for inclusion into the matrix.

Laboratory Tracking -- All medical facilities jea a mechanism for tracking laboratory data. If laboratory procedures are done in offices, quality standards need to be met. Focused tracking can make significant input into medical care process and outcome analysis.

Encounter Data -- Medical care sites can use encounter data for multiple inputs into the matrix. Tracking the percentage of visits for chronic illness or multiple-diagnoses-patients begins to determine the severity of patient mix. Tracking the percentage of psychological visits begins to measure the medical facility's attentiveness to these issues.

Software Prototype

In the first software package prototype for the proposed matrix, each square is measured on the basis of a 100-point maximum score. Customizing the input into each square is not difficult to build into the software. There is an ability to track individual doctors, individual sites, isolated specialties, and facilities as a whole. The program allows for the easy addition or deletion of doctors, facilities, and specialties. The present prototype is IBM-compatible and has a data security system to prevent inappropriate access. Every attempt has been made to make this program user friendly. Its basic operating environment is MS-DOS. There are impressive graphic built-ins to ease comprehension of data analysis. It can assemble questionnaires for survey usage and can accommodate manual as well as programmed data collection. It is available in both 5 1/4" and 3 1/2" disk format and requires approximately 1.5 megabytes of disk storage and 640K RAM memory.

[1] Horn, S. "Ambulatory Severity Index: Development of an Ambulatory Care Mix System." Journal of Ambulatory Care Management :53-62, Nov. 1988.

[2] Craddick, J. "Severity of Illness Adjustment Systems." Quality Resource Monitor .4-5, Sept.-Oct. 1987.

[3] Mohlie, S. Presentation at Group Health Association of America Conference, Aug. 26-27, 1988, Toronto, Ontario, Canada.

Raymond J. Fabius, MD, is Medical Director, CIGNA Healthplan of Pennsylvania, and President, Drexel Hill Pediatric Associates, Philadelphia. He is seeking grant and/or investment capital for further development of this ambulatory care quality measurement package. If successful, he plans to publish a report evaluating its usage in a multitude of medical settings.