SUMMARY

The National Water Quality Laboratory (NWQL) has implemented new procedures for establishing reporting levels and for reporting low-concentration data that are produced by inorganic and organic methods. Advantages of this implementation include

• A statistically based approach for the determination of long-term method detection levels (LT–MDL’s) that is applied uniformly to most NWQL methods and is determined by using routine production-laboratory conditions.
  
• The use of the LT–MDL as the lowest reported concentration provided by most methods to help minimize the risk of reporting a false positive (analyte reported present when it actually is not in the sample). The risk of reporting a false positive that has a concentration equal to or greater than the LT–MDL concentration should be no more than 1 percent.
  
• The reporting of estimated concentrations below the LT–MDL for detected analytes determined by organic methods, such as mass spectrometric or photodiode array ultraviolet/visible spectroscopic detection, that are classified by NWQL as “information-rich methods.”
  
• Yearly re-determinations of LT–MDL’s to provide a continuous assessment and updating of method detection capability at low concentration.
  
• Availability of performance information (mean, standard deviation, and 95-percent confidence interval of mean, percent recovered) for low-concentration quality-control spike samples that are used to determine the LT–MDL.
  
• A documented procedure for establishing the default “less than” reporting level (called the laboratory reporting level [LRL]) that is calculated by using the LT–MDL. [LRL = 2 x LT–MDL]
  
• A minimized risk of reporting a false negative at the LRL. No more than 1-percent probability of reporting that the analyte is not present when it actually is present at or above the LRL concentration. If the true concentration of the analyte in the sample is equivalent to the LRL concentration, the analyte should be detected and reported 99 percent of the time, although the reported concentration might be greater or less than the true concentration.
  
• Data-reporting conventions that provide more low-concentration data. Estimated (“E” coded) concentrations are provided for analytes detected in the region from the LRL or lowest calibration standard (LS), whichever is greater, down to the LT–MDL.

Some LT–MDL’s and LRL’s will change over time. Reporting-level changes are scheduled to take effect at the beginning of the water year (October 1).

To interpret data reported under the new conventions, users must know the applicable LRL, LT–MDL, and in some cases, the LS, in addition to the reported concentration. Until this information can be stored with the analyte concentration in NWIS, it will be available to data users at the NWQL web page http://wwwnwql.cr.usgs.gov. Under the “Technical Information” section on this web page, click on “Long-Term Method Detection Level and Laboratory Reporting Level Information.”

Adoption of the LT–MDL process places responsibility for using and presenting final data with the user rather than with the laboratory and results in substantial interpretive benefits to the data user. Decisions about whether the intended use of the data merits censoring can be made by the user, although uncensored data are available for interpretation and statistical analysis if that is the intended use. Informed use of these data requires knowledge of how the data will be used, what the anticipated future uses of the data are, and how data certainty affects intended data interpretations.

The effect of this change in reporting conventions on interpretations when using current and historical data depends on whether the new LRL is less than or close to the historical minimum reporting level (MRL) or is substantially greater than the historical MRL. For the former outcome, no special action is required. The latter outcome requires more caution as historical and current data are combined and interpreted because the higher new LRL might indicate that the previous MRL was set too low. Thus, historical data reported below the new LRL might be considered estimated values at a minimum. The user may choose to censor historical data in data reports at the new LRL when a high degree of certainty is required. These are actions that must be applied to the data by the user after retrieving historical data from the data base. No changes will be made to data already stored in the data base based on new LRL’s.