analysis, as discussed in the first article in this series, Outcome Analysis for Vascular Access Devices: System Development,! has become an important component of clinical practice. After defining the data for a technology assessment storage program, the structure and content of that data must be considered. Information must be gathered in a systematic fashion that is consistent and descriptive. Ideally, this information could be gathered at multiple facilities, enhancing the power of the information by eliminating single institution bias. Unfortunately, if the data are not described in the same way, no collective conclusions or comparisons can be drawn. A large amount of information could be available, but the result has been a classic apples and oranges predicament, in which each team describes the same problem, but with terminology different enough to prevent meaningful combination of the data sets. A clinical example of consistent data description is a cancer staging system, such as that published by the AJCC (American Joint Committee on Cancer).2 If a woman is diagnosed with Stage IIa breast cancer, possible treatments could be identified from the experience of cancer centers worldwide, all of whom use the same staging system. In performing such a literature search, treatments reported for any other stage breast cancer need not be examined as closely, as the consistency of the reporting system identifies relevant information. This consistent approach to defining the clinical extent of the malignancy allows caregivers to project the prognosis, or probable outcome, of a patient's disease state. Similarly, to accurately analyze the impact of central venous access devices on patient care, a formal vocabulary must be established to permit accurate and consistent description of clinical parameters that may affect the function, complication rate, or longevity of the VADs. This vocabulary will be the first step in ensuring the quality of the information in a VAD outcome database. Many elements of the database discussed previously have obvious descriptors. For example, when describing the vein used for the catheter insertion, there are a limited number of choices (e.g., subclavian, internal jugular, external jugular, cephalic, femoral, etc.) for an even more limited number of sides (right or left). The resulting description of the venous access site will be consistent across a broad range of clinical institutions. Another example of a straightforward data parameter is the reason for VAD removal. Generally speaking, almost all VADs are removed for one or more of the following reasons: end of treatment, infection, thrombosis, malfunction, accidental dislodgement, or catheter compression/fracture. Other reasons are possible, but very infrequent.