Currently, medical decisions mainly are based on results of laboratory diagnostic tests, which are used to confirm, exclude, classify or monitor disease in order to guide treatment. Therefore, establishing appropriate reference ranges is crucial for the correct interpretation of laboratory results. Unfortunately, our observations show that the main source of reference intervals in Poland is data from literature, often based on research carried out among the general population or with a different genetic and/or cultural profile [14]. The second source of information about the expected values is provided by the laboratory reagent manufacturer. The analysis of the RI values in these materials shows that although the manufacturers declare that their procedures comply with the CLSI recommendations, the reference groups are very often too small in number and very poorly characterized in many important aspects, such as race, disease, or body weight [15–17]. Very rarely, a laboratory procedure is also used in order to verify the reference intervals provided by the manufacturer. This approaches involve a comparison of the values of the data approved by CLSI with a laboratory reference group of at least 20 people, which is a substitute for the standard method based on the analysis of the results of at least 120 people accepted as reference [1]. Considering the above-mentioned common weaknesses or even shortcomings in determining reference intervals, indirect methods and their possibility of implementing computer applications [18], may will be the best alternative for regional hospitals and field laboratories serving the same population of a given region. This is especially useful in terms of accessing hard-to-reach paediatric or geriatric populations.
There are a limited number of publications available using the Hoffman for RIs estimation in the general population, and particularly in the paediatric population. Initiatives are being taken around the world to harmonize paediatric reference intervals to improve paediatric diagnostics with a range of effects from national to global. Reported thyroid hormone RIs in childhood vary widely and knowledge on the matter is sparse. Reported RIs are mainly based on direct methods, which, combined with limited availability to the child population, makes them difficult to implement for the average laboratory [19]. We showed that paediatric values of RI for TSH were higher than those provided by our reagents manufacturer, and the greatest differences were observed for LRI (Table 2). The literature overview indicated, the lower limit for TSH ranged between 0.32 and 1.30 mIU/L in children aged ≥ 1 years, and the highest value is similar to the value observed in our study for children aged ≥ 1 (LRI for children ≥ 1 y. <6: 1.03 mIU/L and 0.79 mIU/L for children ≥ 12 y. <18). The upper limit for TSH in an overview by Onsesveren et al. [9] ranged from 2.36 to 6.57 mIU/L, as reviewed by their studies.
In CALLIPER project the RI values determined by the direct CLSI method were compared with the values determined by the Hoffman method showed a much wider range of RIs, determined by the Hoffman method for TSH and fT4. This study indicate the limited usefulness of the values determined by the Hoffman method, especially if it is carried out on a limited number of results from centres providing highly specialized care for children, which may cause a large percentage of pathological results [20]. The values determined in our study by the indirect RI method for TSH were very similar to those obtained in the CALIPER study, although direct comparison is very difficult due to the use of different age ranges in the CALIPER study for the three different immunochemical analysers applied. However, in both studies a significant TSH URL decrease was observed in participants aged 12–14 years to values similar to those of adults. As highlighted by the authors themselves, establishing assay-specific reference intervals for immunoassays is a need of great importance [21]. The observed differences may be due to the recognition of different epitopes of TSH by monoclonal antibodies used in various kits and microheterogeneity TSH molecules [22]. In our opinion, this approach is far more feasible when using the indirect method on a hospitalised population.
An estimation of TSH RIs for a Mediterranean population over 15 years old was performed by Lo Sasso et al. [23]. The TSH RIs estimated in this study, which had enrolled over 22 602 individuals, were 0.18–3.54 mIU/L in the general population – 0.19–3.23 mIU/L in men and 0.18–3.94 mIU/L in women. Moreover, their results showed a significant interaction between sex and age, suggesting that the effect of age on TSH between the sexes. However, this comparison was based on a statistical test, not on RCV like our study. Our results in the adult population indicated significantly higher values of LRI and much more similar values for URI. This difference may be due to the use of a different reagent kit and analyser in Lo Sasso's et al. [23] study, for which the RIs values provided by the reagent manufacturer were significantly lower than for the Siemens kit used in our study (0.20–4.20 mIU / L Cobas e801 vs. 0.55–4.78 mIU / L Atellica IM). This fact again indicates the necessity to determine the assay-specific RIs for the TSH. Moreover, our study also confirmed the trend observed by Lo Sasso et al. [23] of higher URI for adult women and a decrease of both LRI among participants from 18 to < 90 years old. Similar results to ours were also obtained among the Turkish adult population by Inal et al. [7] from results of TSH and fT4 from hospitals and health centres; data were prepared only by logarithmic transformation, with repeated exclusion of outliers and computation of non-parametric 2.5th and 97.5th percentiles. Indirect RIs were then compared with RIs established by the IFCC direct method (RIs currently used in Inal laboratory) and manufacturers' recommendations. This comparison revealed the largest difference in LRI, where the manufacturer’s value was 1.6 mIU/L lower than established by the indirect method and 1.3 mIU/L lower than estimated through use of the direct method. The differences between direct and indirect methods both for LRI and URI were much less visible, with 0.40 vs. 0.43 for LRI and 3.93 vs. 4.2 for URI, respectively [7].
The study closely related to our research were conducted by Katayev et al. [24] in 2008 on both, female and male adult TSH results collected in 6 laboratories from the Laboratory Corporation of America. Without taking age groups into account, the estimated RIs were in the range of 0.44–3.05 mIU/L for all 129443 results. Comparison of our study to Katayev et al. results [24], the greatest differences were also seen for the LRI, and ranged from 0.19 mIU/L (30.1%) for the ≥ 18 y < 40 group to 0.05 mIU/L (-12.8%) for the ≥ 90 years. The higest absolute difference (2.15 mIU/L) was observed for URI in the ≥ 90 age group, but the percentage difference was as high as 41.3%. Even though Katayev et al. [24] found a substantial (though still below the RCV) difference between their RIs and the literature data, they emphasised that the proper use of statistical techniques and the large number of observations allowed to estimate RIs with accuracy using the indirect method.
Apart from the TSH tests, the second most important parameter in the diagnosis of thyroid disorders is the concentration of fT4 [25]. As we indicated above, the literature data shows large differences in RIs regardless of the method used for its estimation. According to a literature overview by Onsesveren et al. [9], the LRIs for fT4 range from 0.54 to > 0.78 ng/dL (from7.0 to > 10 pmol/L), while the URIs vary from 1.20 to > 2.33 ng/dL (from 15.5 to > 30.0 pmol/L). In the same paper, the authors presented RIs derived from their prospective study – 1.07–1.62 ng/dL (13.8–20.8 pmo/L) assessed as 2.5th and 97.5th results obtained among 4273 thyroid disorder-free children at a median age of 6 years [9].
The CALIPER study defined the childhood fT4 RIs as 0.89–1.70 ng/dL for infants and 0.89–1.37 ng/dL for children under 19 years by Abbott ARCHITECT analyser [21] – which figures similar to results obtained by our indirect method and using another analyser.
The fT4 RIs discussed above take into account values obtained by the direct method and are generally consistent with our results, even though they are potentially influenced by many factors, such as weight, sex, ethnicity, and the laboratory method of determination. However, fT4 RI values among children derived from the indirect method are absent in literature. The most similar study was published by Kapelari et al. [26] for the Austrian child population. In this study, a posteriori direct methods were applied to child hospital populations in five subsequent age groups. Authors concluded that the variation of fT4 concentration was the highest in first months of life and then RIs values decreased in a continuous fashion. LRI were lower in infants, at 0.71 ng/dL (9.17 pmol/L), and then increased up to about 0.82 ng/dL (10.6 pmol/L) in older age groups. In our study we observed similar values, but there were no real differences in values. We noticed the same trend for URI (1.61 ng/dl for infants and 1.33 ng/dl for adolescents). This was generally consistent with the Kapelari et al. [26] study, although their values were higher: 1.97 ng/dL for infants and 1.78 ng/dL for adolescents. For adults, fT4 RIs were estimated using the direct method in euthyroid populations; 0.78–1.55 ng/dL (10–20 pmol/L) were found in a British study [5] and 0.70–1.56 ng/dl (12.29–20.03 pmol/L) among the population of the Republic of Srpska [27]. These results are generally in compliance with those established in our study (with minimal higher values) and data from the manufacturers of the reagents we used (minimal lower values). Moreover, our results corresponded with sparse studies using the indirect method, such as the study by Jakubowski et al. [28] in Poland, which was conducted over 16 years. In this study, the authors applied many exclusion criteria linked to the pre-analytical phase and clinical factors that suggested severe or endocrinological disease and to testing the same patient more than once. The 2.5th and 97.5th percentile of that prepared data were adopted as the low and upper RIs for fT4: 0.86–1.23 ng/dl (11.13–15.87 pmol/L) and were narrower than provided by their reagent manufactures at 9.01–19.05 pmol/L. The results of ours study indicated a substantial shift in URI to higher values than in Jakubowski et al. [28].
The determination of reference intervals for thyroid hormones is a subject of ongoing debate due to poor standardisation of immunochemical methods and the lack of unambiguous reference materials [29] as well as large inter-population differences, along with variability related to the age and gender of patients within the same population. The legitimacy of determining reference intervals for thyroid hormones with an indirect method based on the results of hospital tests is additionally supported by the fact that these tests are screened, and therefore a large portion of the results comes from people without thyroid disorders. The use of such an approach additionally eliminates the differences in values that may result from different methods of sampling in different hospital wards and collection points [30].
Encouraged by the hints contained in Jones' publication [3], we searched for the values of reference intervals characteristic for the population of our hospital and the methods and apparatus used in our laboratory. At the same time, we do not forget that no intervals are perfect and final, and the results obtained by the indirect method, even if they are not absolutely accurate, are closer to the actual state of the population of a given region, because they take into account the analytical and biological variability of the analysed parameter.