RAND analyzed 2005 Medicare facility data for services paid through the hospital OPPS. The data file contained summary data aggregated to the diagnosis-service category level. This level of detail provides sufficient information to understand, in the aggregate, the types of services Medicare beneficiaries receive, but lacks specificity to describe individual patient encounters or episodes of care. CMS provided two data files, which included the diagnosis for an encounter,10 as well as visits aggregated to the Ambulatory Payment Classification (APC)11 level or the Healthcare Common Procedure Coding System (HCPCS) level. Each file contained code descriptions (APC, HCPCS, International Classification of Disease Version 9 or “ICD-9”), the total frequency, the APC paid in 2005, the 2007 payment rate for either the APC or HCPCS (total, and by diagnosis), and a CMS status indicator describing the type of service.
These data were analyzed to determine the following:
- The most common reasons (diagnoses) for visits (E&M services),
- The most frequent services/procedures provided,
- The services/procedures representing the largest costs within this setting,12 and
- The most frequent drugs and biologicals provided in this setting.
E&M visits were identified using the status indicator V (i.e., the status indicator associated with APC codes that indicate clinic or emergency department visits). Services/ procedures were identified with the status indicators S, T or X (i.e., the status indicators associated with APC codes that indicate significant services/procedures and ancillary services). Drugs and biologicals were identified using status indicators G (pass through drugs and biologicals), H (pass through devices, radiopharmaceuticals, brachytherapy), or K (non-pass through drugs and biologicals)13. The analyses did not include laboratory services14 or durable medical equipment (DME),15 which are not paid under OPPS.16
The total cost associated with the provision of each service/procedure was calculated by multiplying the frequency of the service/procedure by the 2007 APC payment for that service/procedure to obtain total Medicare costs. In our analyses, we applied 2007 payment rates to the 2005 utilization data; therefore, the estimates of 2007 spending based on these calculations assume that the volume and distribution of visits and services/procedures did not substantially change over the two year period.
Under Medicare OPPS rules, multiple APCs may be reported on a single claim when patients receive multiple, separately billable services. For example, a patient visiting the HOPS may be billed for a clinic visit (an E&M-related service), a chest x-ray, and an electrocardiogram during the same encounter. Because the files we used for these analyses did not have patient- or encounter-specific data, we were unable to explicitly link visit data (i.e., APCs with status indicator V) with significant services/procedures (i.e., APCs with status indicator S, T or X). Therefore, we cannot describe the spectrum of individual services a Medicare beneficiary receives during a single visit (e.g., we could not identify at the patient level, multiple services/procedures as part of the same encounter, or patients with E&M services/procedures during the same encounter).
For each common or costly APC representing services/procedures, clinical experts at RAND identified the specialties that most frequently bill for these professional services based on data from the American Medical Association’s (AMA) 2005 Relative Value Scale Update Committee (RUC) database. This database indicates the specialties that commonly bill for individual services/procedures at the HCPCS (Current Procedural Terminology [CPT]) level. In making the determination, RAND examined the providing specialties for any HCPCS code that accounted for at least five percent of the claims within an individual APC in 2005. This assignment was done to assist in the identification of measures potentially relevant to common services delivered in the hospital outpatient setting.
To facilitate examination of diagnoses associated with visits and services/procedures, RAND researchers grouped common diagnoses. Individual diagnoses were aggregated into diagnostic groups by two physicians using headers in the ICD-9-CM codebook as a guide.17 Diagnoses were also grouped by organ or body systems. The main driver for grouping diagnoses was to ensure that the most common diagnoses that have multiple diagnosis codes at the four-digit level (e.g., diabetes, hypertension) were aggregated, thereby allowing our analyses to accurately reflect their collective frequency and costs.18
We examined E&M visits separately from services/procedures to assist us in our efforts to identify performance measures, as E&M visits mimic the type of preventive, acute and chronic care provided in the ambulatory setting for which a large number of measures currently exist. Additionally, all data analyses were performed separately for the ED and the HOPS, given the distinct types of care provided by these two departments.