The Honorable Vera Katz

The Honorable Vera Katz
Speaker of the House
State Capitol
Salem, Oregon 97310

Dear Representative Katz:

As you requested in our meeting May 7th, the Department of Insurance and Finance has examined SAIF's medical-cost data to address questions of proper interpretation and data quality. This data came from SAIF's detailed 1988 and 1989 provider payments reports (CZPO12-B) and summary "Medical Costs by Provider Type: 1988"

Dr. Nyiendo's criticism of the use made of 1988 data is essentially valid: The data did not fairly compare actual medical costs among providers. The format used by SAIF was not suited for comparing average costs paid to one type of medical provider with another. Further, corrections to the detailed data which SAIF made before compiling the 1988 summary cannot be verified. SAIF made manual adjustments to the data before summarizing it, but did not keep a record of the changes.

After reviewing potential problems identified by Dr. Nyiendo for payments to specific providers, we concluded that the payments or adjustments were appropriate. But, the way these dollars were categorized by provider-type was sometimes wrong. Even when the dollars were assigned to the correct category, the way claims were counted was inaccurate. For example, the listing of general merchants (such as K-Mart and Sears) categorized as "medical physician" turned out to be appropriate payments which should have4 fallen into the "clothing (voc)" group and should not have been counted medical payments at all.

The negative amounts which appeared on the report were appropriate. They were adjustments which corrected for payments made in error or to reflect reimbursements from the department for certain types of claims costs. The corresponding claims counts, however, were not corrected. It appears that when the wrong party was paid on a claim and the money was later recovered, the claims counts were never adjusted. Providers had incorrect zero claim counts if they supplied no tax identification number, due to a programming error.

Other line items which looked suspicious were legitimate. For example, the payment to Wildlife Safari turned out to be a case where Wildlife Safari was the employer receiving reimbursement for drugs which they had purchased for their injured worker. The large payment to Learning Services of Boston was for a permanently and totally disabled quadriplegic, injured worker, who is a ward of the court, which determined that he needed these extremely expensive services.

The way SAIF's report CZPO12-B summarizes claims still has problems. However, even if the data were completely clean, the historical summary format is not suited for comparing different provider-types. The format in question is illustrated by the attached summary table, "Medical Costs by Provider Type: 1988." Part of the problem with this table is ambiguous labeling. This table represents the annual payments which SAIF made to medical providers during calendar year 1988, not the total costs of a group of "1988 claims." SAIF provided five summary figures for each type of medical-service provider. These columns are labeled: "Total Cost," "Average Cost Per Claim," "Number of Claims," "Percent of Total Claims," and "Percent of Total Costs." None of these labels may be taken at face value.

The most serious misunderstandings arise from column D, "Number of Claims": It appears from line 31, the "Grand Total," that SAIF paid $87,278,017 for medical service on 193,125 claims at an average cost of $452 per claim. The "Number of Claims" column actually represents the sum of the total number of different claims served by each individual medical provider. For example, an injured worker who went to a chiropractor and then to a medical doctor who was referred to another MD, who was then admitted to a hospital, would be counted as four claims. This one claim would appear in the count for each of the provider types involved and twice for medical doctors. All four counts would be included in "Number of claims." This is a departure from the usual use of the term "claim" to refer to one person's single compensable injury.

We have seen that "total costs" on this table are really annual payments and "claims" are really claims per provider with multiple counting; consequently, the averages and percents based on these figures do not mean what one would assume. We recommend that SAIF correct the erroneous provider-types and rename the "total cost" and "percent of total cost" columns: "annual payments" and "percent of annual payments." In addition, they should delete the claims columns and add explanatory footnotes to minimize the chance of misinterpreting the data. If these actions were taken, this table would present accurate data on the overall distribution of medical payments.

He also examined the 1989 SAIF detailed provider-payments report (CZPO12-B). SAIF has tightened their data-editing since 1988, and the categorization by provider type has probably improved. However, we concluded that some of the same problems still exist, especially regarding how claims are counted.

The essential problem with the medical-cost data distributed by SAIF is that it cannot be tied back to claims information. The detailed listing was apparently designed to produce tax reports, not to analyze medical issues. One cannot compare similar claims, total costs, or services provided from this data.

I don't want to belabor the point, but SAIF's summary data is confusing, easily misinterpreted, and does not tell policymakers the things they need to know. Instead, we need to focus on how medical-cost data should be used as a basis for making public policy. To that end, as I'll discuss separately, DIF is developing a recommended format for all insurers to use in reporting medical data.

Sincerely,

Dan Adelman, Administrator
Information Management Division
Department of Insurance and Finance
Salem, Oregon