Data Use, Quality, Reporting, and Communication
After a general discussion of laboratories, testing data, and reports, this chapter focuses on three topics: “limit” values encountered with laboratory data; Data Quality Objectives (DQOs) and Measurement Quality Objectives (MQOs) for selection of laboratories, methods, and data requirements; and the benefits and issues associated with using uncensored data.
Data censoring and the various limit values in use are traced to statistically defined concepts of critical, detectable, and quantifiable levels set down by Lloyd Currie in 1968. These are discussed in detail using ICP-AES beryllium measurements to illustrate their conceptual and technical subtleties. The current U.S. EPA Method Detection Limit (MDL) receives particular attention due to its pre-eminent position in U.S. regulatory affairs and the legal challenges it currently faces. Recent technical advances, notably those due to ASTM, are also discussed.
DQOs and MQOs involve deliberate quantitative planning of an investigation. These are discussed in terms of the statistical properties required of laboratory measurements, again using simple conceptual examples, for decisions based on individual measurements. DQOs and MQOs for decision-making using entire datasets are also discussed, including whether to use censored or uncensored data and issues involving the actual data distributions that may be encountered. Finally, technical aspects of the use of uncensored data is discussed, as are several non-technical issues related to public relations, risk communication and accreditation requirements.