Comparing Physician-Reported Cancer Management Plans With Medicare Services Received FREE

The National Oncologic PET Registry (NOPR) was developed to provide evidence on the effectiveness of position emission tomography (PET) in the detection, diagnosis, and clinical management of certain cancers, and has been used to inform Medicare coverage policy for PET.^1- 3 The NOPR collects data regarding the intended management strategy reported by physicians before and after PET, with a change in intended management suggesting “added clinical value” from the PET scan.⁴ The NOPR does not collect data on the management plan actually implemented. To determine whether physician-reported management plans in NOPR reflect actual services received, we compared the physician-reported NOPR management plan with services provided according to Medicare claims data. As the first study, to our knowledge, using linked NOPR and Medicare data, this comparison provides information on whether physician-reported data in NOPR reflect the care that is ultimately provided to patients.

We conducted a retrospective cohort study using linked cancer registry, NOPR, and Medicare data for Medicare patients in North Carolina and California who received NOPR-documented PET scans from May 2006 through December 2008. Patients with a new primary diagnosis of pancreatic adenocarcinoma or renal cell carcinoma from 2003 through 2007 were identified. Of the 683 eligible PET scans for these cancer cases, we excluded 194 (28.4%) because of nonconsent, noncontinuous Medicare fee-for-service coverage, or death within 30 days of PET scan.

From the post-PET NOPR survey we identified the referring physician's planned management following the PET scan. Using Medicare claims data, we identified management services received in the 30 days following the PET scan based on all diagnosis and procedure codes found in claims. For the NOPR vs Medicare comparison, we categorized management strategies as observation, additional imaging, biopsy, or treatment (eg, surgery, chemotherapy, and/or radiation). In the Medicare data, observation was defined as having no subsequent claims for cancer-related imaging, biopsy, or treatment in the follow-up period.

We assessed the agreement between post-PET NOPR physician-reported intended management and post-PET Medicare services using Cohen κ and calculated the proportion of NOPR post-PET intended management strategies provided based on Medicare data. Since the NOPR allows the referring physician to select 1 intended management strategy (observation, additional imaging, biopsy, or treatment) yet multiple treatments, we also created a variable indicating any treatment.

We identified 489 PET scans for 325 patients. Agreement of post-PET management strategies in NOPR and Medicare services received ranged from poor (κ = 0.06) for additional imaging to fair (κ = 0.49) for surgery (Table). In 21.3% of scans, NOPR physicians indicated planned subsequent treatments that were not found in Medicare claims. Conversely, 11.3% of scans had Medicare claims for treatments after the PET that were not indicated in NOPR as the intended management strategy.

The proportion of NOPR-reported management plans that could be matched to a corresponding Medicare claim ranged from 36.7% (additional imaging) to 72.7% (observation) depending on the management category. For NOPR-reported management strategies that included an intended treatment, 49.0% of surgery, 50.0% of chemotherapy, and 45.5% of radiation therapy were also found in Medicare claims. The results of sensitivity analyses changing the window in which Medicare claims were included (30-60 days after PET, or inclusion of 15 days prior to PET) showed no significant effect on the agreement.

Prior studies examining changes in the physician-reported intended management plans on the pre-PET vs post-PET NOPR forms demonstrated that a significant percentage of intended management plans change after the PET scan.^5- 7 However, to our knowledge, physician reports in NOPR have not been previously validated. In this pilot study, we found only modest agreement between the post-PET management plan indicated by physicians in NOPR and health care services captured in Medicare claims. This finding has important implications for Medicare's Coverage with Evidence Development policies⁸ and others who examine NOPR data to understand PET scan effectiveness for cancer care.

Oncologic care is increasingly multidisciplinary, and treatment plans are often determined or revised after a patient has consulted with several physicians, yet only 1 physician completes the NOPR forms. As such, one must consider that the physician ordering the PET scan may not know the exact management plan before or after the scan. Furthermore, the NOPR physician-reported management strategies are recommendations, not reports of actions taken. As the clinician and patient discuss care options, they may reach a joint decision regarding future care, informed by the patient's individual situation and preferences, that differs from that reported to NOPR.

By examining the relationship between intended and actual management, we demonstrate that post-PET physician-reported management plans in NOPR are frequently not implemented. These results suggest that the impact of PET scans cannot be assessed using the NOPR alone and that supplementation of NOPR with Medicare data is needed to fully assess the clinical impact of PET scans in cancer care.