With demand for endocrine tests steadily increasing year-on-year, review of requesting behaviour is very pertinent. A wealth of data now exists to suggest that there is a significant degree of unnecessary requesting of pathology tests,1-3 including endocrine tests, with commonly requested laboratory investigations imposing a major cost burden upon healthcare systems. We recently determined the re-testing intervals in patients treated with levothyroxine in two centres, in order to compare observed monitoring frequency (re-testing interval) to best practice and to assess the effect of the initial thyroid function test (TFT; TSH, thyroid-stimulating hormone and fT4, free thyroxine) results and the source of the request on the TFT re-testing interval. All TFTs performed by the Clinical Biochemistry Departments at the Salford Royal Hospital (2009–2012; 288 263 requests from 139 793 patients) and University Hospital of North Midlands (2011–2014; 579 156 requests from 193 035 patients) were extracted from the laboratory computer systems. Of these, 54 894 tests were on 13 297 patients confirmed to be on levothyroxine therapy in the test cohort (Salford) and 67 298 requests on 11 971 patients in the confirmatory cohort (North Midlands). In the test cohort, median TFT re-testing interval in the total group was 19.1 weeks (IQR 9.1–37.7 weeks), with clearly defined peaks in TFT re-testing evident at 6 and 12 months and a prominent broad peak at 1–3 months. Median re-test interval was much lower than recommended (52 weeks) for those with normal TFTs at 31.3 weeks (30.6 weeks for the confirmatory cohort). Where TSH was elevated and fT4 was below the reference range, re-test interval was much longer than is recommended (8 weeks) at 13.4–17.6 weeks (7.1–23.4 weeks in the confirmatory cohort), as was the interval when TSH was below and fT4 was above the normal range, at 16.7–25.6 weeks (27.5–31.9 weeks in the confirmatory cohort). There was no observable difference in the pattern of thyroid function testing by age band (<30 years, 30–65 years and >65 years of age). In those cases with initial normal range TFTs, 72.4% (North Midlands: 76.9%) of TFT tests were requested prior to 11 months after a normal TFT result (and 13.2% after 13 months; North Midlands: 11.6%). Overall, in those cases with initial TFT results outside the laboratory reference range, 60.3% (North Midlands: 58.0%) of tests were requested after 10 weeks and prior to 11 months (and 18.0% prior to 6 weeks; North Midlands: 9.1%). Only 21.1% (North Midlands: 11.7%) were requested at the recommended 8 weeks interval (± 2 weeks). There was significant within-practice variation in test-retest interval. For normal range TFTs, some practices showed re-test intervals ranging from <10 weeks to >80 weeks. In those cases with abnormal initial TSH and/or fT4, the median interval was above the recommended 8 weeks in most practices, with between-practice median intervals ranging 8–25 weeks (3.1-fold variation). Again, within-practice variation was considerable, ranging from <5 to >50 weeks in some cases. Relative to recommended monitoring intervals, testing frequency tended to be too short for those with normal initial TFTs and too long for those with TSH and/or fT4 outside the reference range. The finding that 72.4% of cases with initial TFT results within the laboratory reference ranges have repeat tests that are too frequent (<11 months) mirror our findings in patients with diabetes requiring regular HbA1c checks for glycaemic monitoring, that in those with a relatively well-controlled HbA1c (<7%; <53 mmol/mol) 21% of tests were repeated too soon and 30% too late.4 We observed significant within-practice variability in time interval to repeat TFTs in patients taking levothyroxine, with a tendency to repeat TFTs (over-test) in patients with normal initial TSH and fT4 levels, and under-test those with abnormal TFTs. We propose that direct requesting from the clinical laboratory (with a facility for clinician over-ride) may bring patients more quickly to target with their TSH levels and reduce costs in relation to unnecessary testing of TFTs when patients are already biochemically euthyroid. Such a change in practice, while requiring thorough evaluation, has the potential to save money and improve patient outcomes. None.
80
