day on which the test is to be performed. It is the responsibility of all testing personnel to ensure this does not occur. Testing personnel must immediately take the appropriate action for all failed QC runs, refer to section 8.4.
In addition, the laboratory's submission states:
To perform this analysis, systematic error was investigated using quality control (QC) data and patient distribution over time. The QC data was reviewed in Levey-Jennings plots and evaluated for bias trends. The patient data was summarized by plotting monthly means, monthly medians and monthly means ("average of normal" or AON) calculated from central 95% of values. In addition, random errors were evaluated by analysis of QC data. Namely, QC data was reviewed to identify >2SD failures and period of elevated CV. Namely, batches of 10 consecutive QC data points were analyzed and the CV was calculated for each QC level during this period. If the CV was >30%, then all patient results run during that time period will be voided. (Ex. E, Tab 1)
The laboratory's response does not include an evaluation of unacceptable QC results or patients affected or potentially affected by the cited QC failures. It is not clear how evaluating aggregate patient (i.e., monthly means and medians) and QC (i.e., bias trends, >2SD, CV) data addressed specific days that the QC was unacceptable and patient results were reported.
The summary (Ex. E, Tab 1) included analytes tested on the Theranos Proprietary Device. We note that general statements such as, but not limited to, were included:
- "Monthly patient distribution variance were noted, but they could not be further substantiated by QC trends"
- "Patient results following a QC failure and during periods of high CV are being voided"
- "Patient distribution and QC trends did not reveal significant trends."
- "Potential biases observed in patient distributions were found to be insignificant relative to the normal reference intervals"
The laboratory's submission did not provide any explanation or documentation to support these statements.
In Ex. E, the laboratory provided a list of patient specimen accession numbers for which the laboratory "identified reports," but did not specify or submit documentation to indicate whether corrected reports were generated and issued.
The laboratory failed to adequately address this deficiency and provide acceptable evidence of correction consisting of the required documentation and information set forth above and in our January 25, 2016 letter.
D5775
The laboratory's allegation of compliance is not credible and evidence of correction not acceptable.
18
