Do Practice Patterns Reflect Practice Guidelines?

Excerpt

The emergence of clinical practice guidelines and consensus statements has stimulated a critical review of the literature and healthy debate among experts regarding how and when to treat patients with various diseases, especially cancer. An article in this issue of Annals of Surgical Oncology by Panigrahi and colleagues comments on the recent medullary thyroid cancer (MTC) guidelines published by the American Thyroid Association.1,2 In using the Surveillance, Epidemiology, and End Results Program (SEER) database, these investigators found that a high number of patients with MTC, treated from 1988 to 2006, received treatment viewed as out of line with the 2009 American Thyroid Association guidelines. Although this may well be true, there are limitations to the conclusions which one can draw from an analysis of such large data repositories. The SEER database and other tumor registries are intended primarily as sources of cancer incidence and mortality rates.3 Information regarding initial treatment is recorded, but follow-up data (except for date and cause of death) such as locoregional recurrences, distant metastases, and subsequent treatments are not routinely collected. Advantages of SEER are as follows: it includes data from large cohorts; it is population-based; it represents all age groups; and it has a wide geographic range, thereby decreasing selection biases that may hamper clinical trials and studies that are based on institutional databases. Information on stage of disease and initial treatment are verified for accuracy; however, specific details regarding the disease and surgical treatment (such as RET mutation status, the exact location of nodal disease, the specific type of neck dissection and completeness of resection), which are typically recorded in institutional, organ-specific databases and clinical trials, may not be included in SEER. The lack of organ-specific details and information on comorbid illnesses and other diagnoses restrict the analyses and conclusions that can be drawn from such large tumor registries. Despite these limitations, one important reason to study population-based registries is that it allows us to examine the management and true effectiveness of therapies performed in the real world, rather than in the tightly controlled settings of clinical trials and high-volume referral centers. Although the use of these large tumor registries and administrative databases (such as Medicare and the Healthcare Costs and Utilization Project) to study patterns of cancer treatment has increased dramatically, one must be aware of the limitations of each database in all aspects of research design (hypothesis generation), data analysis and interpretation, and perhaps most importantly, the development of conclusions and recommendations.4,5 …