Finding blunders in thyroid testing: experience in newborns

We evaluated thyroxin (T4) and thyroid-stimulating hormone (TSH) data along with clinical information from 600,000 newborns. We looked for certain combinations of tests and clinical data that were questionable and possibly mistaken. Our approach suggests that certain combinations of test results, especially the presence of missing results deserved further evaluation for possible blunders. We found that missing tests were frequently the result of oversight. The laboratory used the well-known standard blood-spot-on-filter paper methods for TSH and T4. For quantitation of TSH and T4, we used the time-resolved fluoroimmunoassay available from Perkin Elmer. We found 56 babies with confirmed primary congenital hypothyroidism (PCH) in a total of 600,000 patients. We also found 18 sets of results in the same 600,000 babies that gave inconsistent findings, had missing values, and (or) possible misinterpretations of the clinical and (or) laboratory data. What is an acceptable mistake rate? All mistakes are unacceptable, but there is likely some irreducible mistake rate, and efforts to reduce the mistake or blunder rate still further may not be cost-effective. What can be done is to study the mistake rate per 600,000 babies from year to year; the mistake rate should be decreasing or not changing. This assumes a stable cohort of babies; an assumption that may be acceptable. We applied a form of pattern recognition to identify cases of possible blunders and missing values in either the laboratory or clinical data. What is clear is that we apparently identified some blunders. The 18 mistakes per 600,000 babies may be "very low" and acceptable. We recommend that seeking ever decreasing mistakes is the way to go, and the level of monitoring the data should be very intense given the serious consequences of mis-diagnosed thyroid disorders.