analyzing data) from the International Classification of Diseases (ICD) (U.S. Department of Health and Human Services, 1980). Once the sample of diagnoses is identified, they can be matched with existing medical treatment data from large retrospective medical claims databases (e.g., those of medical insurers or payers) to monitor changes over time in the cost of a treatment episode. Retrospective medical claims data generally provide detailed information, by individual or family unit, including date of visit with the physician or other medical care professional, medical procedures provided, primary and secondary diagnoses, hospitalizations, dispensed medications, and the patient’s and insurer’s payments for each of these services.
The diagnoses selected for pricing by the BLS could be chosen on the basis of the relative amounts spent on the disease so that, for example, heart attacks would be far more likely than conjunctivitis to be chosen, even though conjunctivitis has more entries than acute myocardial infarction in the ICD. When an individual receives treatments for several distinct but co-occurring illnesses, health services researchers frequently either assign each encounter entirely to the primary diagnosis for that encounter or split the utilization and costs equally among the various diagnoses for that encounter.
Given the present state of information technology and medical claims processing, such a procedure could not be implemented in real time for an MCPI that must be published within a month of data collection. But currently it is feasible with about a 1-year lag, allowing the BLS to publish an experimental index on an annual basis.
Because the BLS would probably find it too costly to produce this type of medical care price index each month, we recommend it consider producing an index based on a pricing of the sample of treatment episodes of distinct diagnoses. These diagnoses have two components—quantity weights and prices, just as with any other price index. For the monthly medical care index, the BLS could reprice the current set of specific items (e.g., anesthesia, surgery, medications), keeping quantity weights temporarily fixed. Then, at appropriate intervals, perhaps every year or two, the BLS should reconstruct the medical care index by pricing the treatment episodes of the 15 to 40 diagnoses in the manner described above.
The index, therefore, might have a break when the set of diagnoses are repriced. There is the possibility that it will jump (up or down), particularly for medical conditions whose treatments have undergone rapid technological change. The panel recommends that BLS explore the possibility that these breaks may be large. A research program could retrospectively estimate the magnitude of such changes and determine what should be done (for example, smoothing the quantity changes based on past trends) if the breaks are expected to be large. At this point, we are agnostic concerning the most appropriate procedures and recommend that the BLS form a study group to examine these issues.