| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
1 Clinical Pathology Department, Clinical Chemistry and, Hematology Laboratory, San Bortolo Hospital, Vicenza, Italy
2 Clinical Chemistry and, Hematology Laboratory, Hospital of Verona, Verona, Italy
aAddress correspondence to this author at: Clinical Chemistry and Hematology Laboratory, San Bortolo Hospital, Via Rodolfi 37, 36100, Vicenza, Italy. Fax 39-0444-752501; e-mail davide.giavarina{at}ulssvicenza.it.
To the Editor:
Grossi et al. (1) recently reported an interesting project that used a sophisticated algorithm for the formulation of reference intervals based on 
15 000 000 records related to 197 350 individuals. We noted an important difference between their reference interval calculated for thyrotropin (TSH) based on results obtained with the Architect (Abbott) analyzer in women (0.28–4.45 mIU/L) and that recently reported by Kratsch et al. (2). Kratsch et al. selected a group of 870 blood donors with negative thyroid ultrasonography and thyroid autoantibodies, as recommended by criteria of the National Academy of Clinical Biochemistry, and found a reference interval of 0.4–3.77 mIU/L (2). The optimal serum TSH reference interval is strongly debated, and a lowering of the upper reference limit is advocated by some authors (3). Furthermore, the algorithm used by Grossi et al. (1) cannot be implemented easily in most institutions because it requires considerable hardware and software resources and statistical expertise that are not commonly available. In our opinion, indirect methods are much simpler and more practical tools for the calculation of reference values or health-related limits (HRLs), especially when the fraction of pathologic values is not too high (4)(5)(6).
We retrieved the results of thyroid panels (which included measurement of anti-thyroid peroxidase antibodies) from the records of 15 359 female and 3862 male patients stored in our laboratory information system (LIS) over a 30-month period (January 1, 2003, to June 26, 2005). We calculated the Advia Centaur (Bayer) TSH HRL, using the program GraphROCTM (Fig. 1A
), and obtained an upper limit of 3.7 mIU/L. As shown in Fig. 1B
, the upper limit of the HRL did not change substantially after we removed the repeat tests (2893) and the results obtained in individuals positive for thyroid antibodies (7995).
|
This limit confirms a previous study carried out in 2000 with the same analyzer and the same software in 40 095 and 26 001 results retrieved from the LIS of the laboratories of Vicenza and Verona hospitals, without any selection criteria; the HRLs were 0.28–3.5 and 0.22–3.6 mIU/L, respectively, and the test result distribution appeared unimodal (7).
The 3.7 mIU/L limit is also consistent with those reported in 2 multicenter studies carried out in Spain [144 reference individuals (8)] and in the United Kingdom [303 individuals (9)] with the same analyzer; the reference intervals obtained in those studies were 0.43–3.69 (8) and 0.48–3.63 mIU/L (9), respectively. Finally, the 97.5th centile of TSH concentration reported by the National Health and Nutrition Examination Survey (NHANES) III in the decades between 20 and 60 years was between 3.56 and 3.82 mIU/L (10). In conclusion, the retrieval of results from the LIS, followed by a simple treatment and visual inspection of the data, seem to yield results more consistent with clinical requirements than a much more demanding procedure such as the REALAB Project.
In our opinion, the somewhat bimodal distribution of TSH for men and women further reinforces our approach and our conclusion. Although our extraction criterion that a request for anti-thyroid peroxidase antibody testing be included may have led to the selection of an increased number of pathologic results, inducing a certain heterogeneity, we did not use any selection criteria in the study carried out in 2000, and we obtained an unimodal distribution for both the Vicenza and Verona results (7).
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
