PET and PET/CT Reports: Observations from the National Oncologic PET Registry

MATERIALS AND METHODS

The report elements included in our review of PET reports were compiled from the ACR communication guidelines (1), the SNM general imaging guideline (2), and the ACR and SNM PET/CT guidelines (3,4) (Table 1).

A training sample of 20 reports selected randomly from the NOPR database was first reviewed by the 4 physicians on the panel. The reviewers were provided with information from the pre-PET form (8) completed by the referring physician for each case, including the indication for PET, site of proven or suspected cancer (and whether this was pathologically proven), the summary stage, and the pre-PET intended management plan (including treatment type, if that was the plan). Each reviewer graded each report to indicate whether the elements listed in Table 1 were present, not present, or not applicable (if a reviewer judged that a component was appropriately not reported, such as not reporting CT technique if the study was a conventional PET study or not describing the site and intensity of ¹⁸F-FDG uptake if the study findings were stated to be normal). The results of the initial review of the training set were tabulated and discussed in a conference call to achieve a consensus among the reviewers regarding scoring and to refine the list of elements.

Each reviewer was then sent 65 randomly selected reports. Forty-five of these were unique to each reader, and 20 were common across all readers. The reviewers scored the reports in the same manner as had been done during the training phase. The reviews were submitted on an Excel (Microsoft) spreadsheet to the Brown University Center for Statistical Sciences for analysis.

The ratings for the 20 cases reviewed by all reviewers were analyzed to assess the level of agreement among the reviewers. The frequency of yes responses (element present) was tabulated for each component. The degree of similarity among the readers' responses for each component was assessed by the Rand index, which was computed to measure the proportion of agreement between all possible pairs of reviewers (12). The Rand index has a range of 0–1, with 0 indicating no similarity between clusters of responses and 1 indicating perfect agreement. The bootstrap method was used to construct the 95% confidence interval, and a permutation test was conducted to assess whether rater agreement was significantly different from that expected by chance (13).

The results for the 180-case sample made up of the 45 cases unique to each reader were analyzed separately from the 20 cases read by all readers. The frequency and percentage of times each component was present (yes response), as well as the number of times each was not absent (yes or not applicable), were tabulated.

RESULTS

Three cases were excluded from the analysis, 1 from the common cases and 2 from the set of 180, because the reports either were incomplete or did not appear to correspond to the data on indication and location of cancer. There was generally excellent interrater agreement for the cases read in common by the 4 reviewers. For example, 30 of 34 elements had the agreement ≥ 0.70 and about half of these had the agreement ≥ 0.90 across reviewers. On average, the agreement was 0.835 (95% confidence interval, 0.815–0.853) across reviewers and elements. In addition, the level of agreement was significantly different from that expected by chance (P < 0.001). There was poorer interrater agreement, as measured by the Rand index, for the following report elements: reason for study clearly stated (range, 0.59–0.90), clinical indication for study clearly addressed (range, 0.54–0.72), interpretation of findings provided rather than just restatement of findings (range, 0.52–0.80), and differential diagnosis provided (range, 0.47–0.75).

The characteristics of the reports reviewed individually are noted in Table 2. The sample provided a balanced representation of PET studies: reports from 162 interpreting physicians at 146 PET facilities were included in the sample. Twenty-eight percent of the facilities have hospital-based scanners, and 25% have mobile units. The reports span the range of clinical indications: diagnosis, 23.0%; initial staging, 20.8%; treatment monitoring, 15.7%; restaging, 18.0%; and detection of recurrence, 22.5%. More than 30 different pathologically proven (77.5%) or suspected (12.9%) cancer types were included: cancers of unknown primary origin and cancers of the pancreas, stomach, and urinary bladder together accounted for 38%. The summary stage of disease was rather evenly distributed. An alternative form of imaging was the most frequent pre-PET management plan, and observation was the least frequent.

The frequency of reporting of each of the report elements is noted in Table 1. Nine elements were included in more than 90% of the reports for which that element was considered to be applicable: the type or site of cancer; the name of the radiopharmaceutical; the administered activity; the body regions scanned; the date of prior PET or PET/CT studies; the type of prior non-PET studies; the date of prior non-PET studies; the description of the location, extent, and intensity of ¹⁸F-FDG uptake if an abnormality was present; and a correlation of abnormal PET findings with other results (including the CT findings of a PET/CT study). Some elements that were less commonly included but were considered important were a clear statement of the reason for the study (58%); a description of treatment history, if applicable (44%); the date of last treatment (45%); the blood glucose level (44%); the time from ¹⁸F-FDG injection to imaging (52%); a characterization of the CT component for PET/CT as nondiagnostic versus diagnostic (54%); a statement concerning the use of oral contrast material (32%); a statement concerning the use of intravenous contrast material (27%); documentation of comparison with a prior scan or that none was available (30%); and documentation that the study's clinical indication was clearly addressed (56%).

DISCUSSION

The content of the report of an imaging test has important ramifications for how a patient is managed and the study reimbursed. If elements related to the performance of the examination and the relevant findings are not included in the report, information that may be important for appropriate patient management may be lost. The ACR and the SNM have published guidelines describing the relevant elements of a complete report, along with a suggested reporting format (1–4). During the initial quality control of NOPR data, we reviewed several reports in conjunction with the information submitted by the referring physicians on the pre-PET and post-PET forms, in order to check for the internal consistency of submitted data. During this review, it became apparent to us that many elements that were considered important in a report were not being included. We then undertook this review of a random sample of reports submitted to NOPR to document how often the various recommended report elements were and were not included. We found, confirming our initial impression, that many important elements of a PET report were missing in a large fraction of reports, including such items as a clear statement of the reason for the study and a description of previous treatment history. Important technical details often missing or omitted were the blood glucose level, the time from radiopharmaceutical injection to imaging, and the details of the CT technique in PET/CT studies. Several elements may have been available to the interpreting physician and that information may have been used in the interpretation of the study (we could not determine this from the available data), but if these are not documented in the report, information that is potentially important to the referring physician may be lost. Moreover, if the elements are not included in the report, there is no documentation that they were available to the interpreting physician.

Some elements that were included in the review are expected to be in all reports, if applicable, and others are not. For example, all reports should contain the following elements: the reason for the study; type or site of cancer (if applicable); treatment history (if applicable); type of therapy (surgery, radiation therapy, or chemotherapy) and timing of therapy relative to the PET scan; radiopharmaceutical, route of administration, and administered activity; blood glucose level; time from injection to imaging; body region scanned; whether a diagnostic CT scan was performed and, if it was, the type and amount of oral and intravenous contrast material used; whether comparison was made with prior imaging studies; location, extent, and intensity of abnormality (including standardized uptake value, if appropriate); correlation with findings of other imaging studies; incidental CT findings for PET/CT; clinical indication clearly addressed; interpretation of the findings in the report impression (rather than just a restatement of findings); differential diagnosis (if appropriate); and documentation of communication of study findings to referring physician (if appropriate). Some interpreting physicians may not consider each of these elements essential to include in the report. For example, some may not report the type and amount of contrast material used since it is included in their facility's procedure protocol. In general, we consider it important to include this information for comparison of techniques when 2 scans are performed at different facilities. Reports of PET studies done for diagnosis would not be expected to specify the type of cancer and treatment history but should specify the reason cancer is suspected clinically. Not all reports need a differential diagnosis. For example, if a patient has biopsy-proven lung cancer, the scan is done for staging, and no other abnormalities are detected, a differential diagnosis is not needed. Unexpected or urgent findings on the PET scan and incidental findings on the CT scan that may alter management of the patient should be communicated to the referring physician or surrogate. The need to communicate is dependent on the findings and clinical context. For example, in a patient undergoing an initial staging scan for lung cancer, an adrenal metastasis is not an unexpected finding (but may need to be communicated if the patient is known to be scheduled for surgery the next day), but a previously unknown large lytic metastasis of the femoral neck seen on PET and CT images would certainly warrant communication to the referring physician.

Several factors, particularly having the clinical history available (14) and performing a comparison with previous radiologic studies (15,16) at the time of image interpretation, have been reported to increase the accuracy of the interpretations. In a recent review of diagnostic accuracy, Berlin (17) notes that having clinical information and reviewing previous studies decreases the number of misdiagnoses. More accurate reports would be expected to be more helpful to referring physicians and ultimately lead to better patient management. Documenting in a report that clinical information was considered and comparison with prior studies was performed informs the referring physician of key factors considered by the reader.

There are some important aspects of the report that relate and directly link to coding, billing, and compliance (5,18). Payer and coding guidelines state that if the nature of a procedure is not documented in the report, one cannot bill for that procedure. Title XVIII of the Social Security Act, section 1862 (a) (1) (A), allows coverage and payment for services provided to Medicare patients considered medically reasonable and necessary; section 1833 (e) prohibits Medicare payment for any claim that lacks the necessary documentation (in this case, in the PET report) to support the charges on the claim. At a minimum, reporting codes that are not clearly documented in the report could be considered fraudulent if audited for regulatory compliance (5,18).

For PET procedures, CMS has issued specific medical necessity requirements in a series of national coverage determinations, the most recent of which was published on April 3, 2009 (19). The new framework for Medicare coverage of oncologic PET categorizes studies as either for initial treatment strategy (formerly diagnosis and initial staging) or subsequent treatment strategy (formerly treatment monitoring and restaging); later this year, new modifiers will need to be appended to the procedure code to indicate the purpose of the study. The national coverage determination imposes a frequency limit on studies done for initial treatment strategy and requires that some PET studies still must be done under the CMS coverage-with-evidence-development program. An appropriately detailed history that addresses the CMS-defined criteria for initial treatment and subsequent treatment strategies is necessary to determine whether the procedure meets medical necessity for the code billed. Additionally, CMS considers surveillance analogous to screening and not coverable; therefore, it is prudent to document the rationale for PET requested to detect a suspected recurrence in order to demonstrate the medical necessity of the study as a condition of coverage. Lack of detailed information could result in an inappropriate choice of code or modifiers.

Because Medicare has contractual obligations to its beneficiaries and the Medicare Trust Fund, the CMS, as well as other payers, requires reasonable, complete, and legible documentation that services are consistent with coverage policies. Documentation includes clinical descriptions in the report that are consistent with definitions contained in Current Procedural Terminology, International Classification of Diseases and Healthcare Common Procedure Coding System codes and past and present diagnoses that should be accessible and accurately reported to the highest degree of certainty. In Table 1 we have noted the report elements that are necessary for accurate and appropriate coding.

There are several reasons some of the elements may not be in a report. Some of the information that ideally should be included in the report may not have been available to the interpreting physician at the time of dictation (e.g., blood glucose level, time from injection to imaging, and the reason for the study). Frequently a request will just include the type of cancer, but the physician interpreting the study does not know the specific reason for the study. Such a situation suggests a need for the PET center to revise its request initiation process and the types of information the requesting physician should be expected to provide.

CONCLUSION

Several elements that should be in a PET report are not present in most reports. Only 9 of the 34 elements reviewed were found in more than 90% of the reports. These deficiencies may result in reports that are less helpful to referring physicians, cause misdiagnoses, and lead to coding and billing errors. Interpreting physicians should audit their reports to ascertain that they include appropriate elements necessary for billing compliance and for effective communications with referring physicians.