Reference Intervals: Comparison of Calculation Methods and Evaluation of Procedures for Merging Reference Measurements From Two US Medical Centers

Critical appraisal of two Box-Cox formulae for their utility in determining reference intervals by realistic simulation and extensive real-world data analyses

BACKGROUND

The reference interval (RI) is defined as the central 95 % range of reference values (RVs) from healthy individuals. The ideal method for determining RIs is to transform RV distribution into Gaussian and estimate its 95 % range parametrically. One-parameter Box-Cox formula (1pBC) is widely used for correcting skewness (Sk) or kurtosis (Kt) in data distribution. However, 1pBC is not popular for computing RIs due to its unreliability in Gaussian transformation. While its two-parameter version (2pBC) is not used due to a challenge in fitting power (λ) and shift (α) parameters simultaneously. In this study, technical issues in fitting both formulae are assessed, and an alternative algorithm for successful use of 2pBC is proposed.

METHODS

For fitting 1pBC, optimal λ was determined by stepwise linear search. For 2pBC, optimal [λ, α] combination was pursued in two ways: by grid search of λ and α (2pBCgrid) or by using the grid search but keeping α-range close to the reference distribution (2pBCopt). Their accuracy and precision in determining RIs were compared by generating power-normal distributions simulating RVs of 23 major chemistry analytes. Additionally, their practical utilities were compared by analyzing 776 real-world datasets comprising test results of 25 analytes that were obtained from the global multicenter RV study of IFCC. For comparison, the performance of nonparametric method was evaluated in both settings.

RESULTS

For analytes with not-much-skewed distributions, unbiased estimation of RIs was accomplished by all methods. Nevertheless, when reference distributions are located far from zero, λ estimated by1pBC and 2pBCgrid fluctuated widely, which was attributable to virtually flat goodness-of-fit profile for [λ, α]. For highly skewed distributions, 1pBC caused bias in estimating RI and λ due to remotely peaked goodness-of-fit profile. Real-world data analyses revealed that 2pBCopt and 1pBC achieved Gaussian transformation (|Sk|<0.1 and |Kt|<0.3) in 82.4 % and 66.9 % among 776 datasets, respectively. Fitting bias signified by Kt<-0.4 was more common to 1pBC. The practical utility of 2pBCopt was unbiased prediction of analyte-specific distribution-shape (λ). Whereas nonparametric method gave highly variable RIs for both simulated and real-world datasets.

CONCLUSIONS

2pBCopt is suitable for calculating RIs and grasping distribution-shape from the estimated λ.

Determining reference ranges and sample sizes in parallel-group studies

BACKGROUND AND OBJECTIVES

Reference ranges are widely used to locate the major range of the target probability distribution. When future measurements fall outside the reference range, they are classified as atypical and require further investigation. The fundamental principles and statistical properties of reference ranges are closely related to those of tolerance interval procedures. Existing investigations of reference ranges and tolerance intervals mainly devoted to the primitive cases of one- and paired-sample designs. Although reference ranges hold considerable promise for parallel group designs, the corresponding methodological and computational issues for determining reference limits and sample sizes have not been adequately addressed.

METHODS

This paper describes a complete collection of one- and two-sided reference ranges for assessing measurement differences in parallel-group studies that assume variance homogeneity.

RESULTS

The problem of sample size determination for precise reference ranges is also examined under the expected half-width and assurance probability considerations. Unlike the current methods, the suggested sample size criteria explicitly accommodate desired interval width in precise interval estimation.

CONCLUSIONS

Theoretical examinations and empirical assessments are presented to validate the usefulness of the proposed reference range and sample size procedures. To enhance the usages of the recommended techniques in practical applications, computer programs are developed for efficient calculation and exact analysis. A real data example regarding tablet absorption rate and extent is presented to illustrate the suggested assessments between two drug formulations.

Indirect estimation of reference intervals for thyroid parameters using ADVIA Centaur XP analyser

Background

The aim of this study was to determine the reference intervals (RIs) for thyroid stimulating hormone (TSH), free thyroxine (FT4), free triiodothyronine (FT3) and FT3/FT4 ratio using indirect methods.

Methods

We analyzed 1256 results TSH, FT4 and FT3 collected from a laboratory information system between 2017 and 2021. All measurements were performed on a Siemens ADVIA Centaur XP analyzer using the chemiluminescent immunoassay. We calculated the values of the 2.5th and 97.5th percentiles as recommended by the IFCC (CLSI C28-A3).

Results

The RIs derived for TSH, FT4, FT3 and FT3/FT4 ratio were 0.34-4.10 mIU/L, 11.3-20.6 pmol/L, 3.5-6.32 pmol/L and 0.21-0.47, respectively. We found a significant difference between calculated RIs for the TSH and FT4 and those recommended by the manufacturer. Also, FT3 values were significantly higher in the group younger than 30 years relative to the fourth decade (5.26 vs. 5.02, p=0.005), the fifth decade (5.26 vs. 4.94, p=0.001), the sixth decade (5.26 vs. 4.87, p<0.001), the seventh decade (5.26 vs. 4.79, p<0.001) and the group older than 70 years old (5.26 vs. 4.55, p<0.001). Likewise, we found for TSH values and FT3/FT4 ratio a significant difference (p <0.001) between different age groups.

Conclusions

The establishing RIs for the population of the Republic of Srpska were significantly differed from the recommended RIs by the manufacturer for TSH and FT4. Our results encourage other laboratories to develop their own RIs for thyroid parameters by applying CLSI recommendations.

Comparison of reference intervals derived by direct and indirect methods based on compatible datasets obtained in Turkey

BACKGROUND

Indirect derivation of reference intervals (RIs) from the laboratory information system (LIS) has been recently pursued. We aimed at evaluating the accuracy of indirectly predicted RIs compared to the RIs established directly from healthy subjects in the nationwide RI study in Turkey, targeting 25 major chemistry analytes.

METHODS

LIS data were retrieved from the laboratory that performed measurements for the direct study. They were cleaned by limiting to outpatients with age 18-65 years, and by allowing only one record per year per patient. Evaluated were four indirect methods of univariate approach: Hoffmann, Bhattacharya, Arzideh, and Wosniok methods. Power transformation of the LIS dataset was performed either using the power (λ) reported by the IFCC global RI study (the first two methods) or using a λ predicted (the last two).

RESULTS

Compared to the direct study dataset, the LIS dataset showed a variable degree of alterations in peak location and shape. Consequently, lower-side peak-shifts observed in sodium, albumin, etc. led to lowered RI limits, whereas higher-side peak-shift observed in triglyceride, low-density lipoprotein cholesterol, etc. led to raised RI limits. Overall, 72% (62-81) of the RI limits predicted by indirect methods showed significant biases from direct RIs. However, the biases observed in total cholesterol, lactic dehydrogenase, etc. were attributed to a higher-side age-bias in LIS dataset. After excluding them, the overall proportion of biased RIs was reduced to 47% (38-54).

CONCLUSION

To reduce prediction biases that remained after age adjustment, it is necessary to apply more rigorous data-cleaning before applying indirect methods.

Big data and reference intervals: rationale, current practices, harmonization and standardization prerequisites and future perspectives of indirect determination of reference intervals using routine data

Reference intervals are commonly used as a decision-making tool. In this review, we provide an overview on "big data" and reference intervals, describing the rationale, current practices including statistical methods, essential prerequisites concerning data quality, including harmonization and standardization, and future perspectives of the indirect determination of reference intervals using routine laboratory data.

Establishing Ghanaian adult reference intervals for hematological parameters controlling for latent anemia and inflammation

Background

In Ghana, diagnostic laboratories rely on reference intervals (RIs) provided by manufacturers of laboratory analyzers which may not be appropriate. This study aimed to establish RIs for hematological parameters in adult Ghanaian population.

Methods

This cross‐sectional study recruited 501 apparently healthy adults from two major urban areas in Ghana based on the protocol by IFCC Committee for Reference Intervals and Decision Limits. Whole blood was tested for complete blood count (CBC) by Sysmex XN‐1000 analyzer, sera were tested for iron and ferritin by Beckman‐Coulter/AU480, for transferrin, vitamin‐B12, and folate was measured by Centaur‐XP/Siemen. Partitioning of reference values by sex and age was guided by “effect size” of between‐subgroup differences defined as standard deviation ratio (SDR) based on ANOVA. RIs were derived using parametric method with application of latent abnormal values exclusion method (LAVE), a multifaceted method of detecting subjects with abnormal results in related parameters.

Results

Using SDR ≥ 0.4 as a threshold, RIs were partitioned by sex for platelet, erythrocyte parameters except mean corpuscular constants, and iron markers. Application of LAVE had prominent effect on RIs for majority of erythrocyte and iron parameters. Global comparison of Ghanaian RIs revealed lower‐side shift of RIs for leukocyte and neutrophil counts, female hemoglobin and male platelet count, especially compared to non‐African countries.

Conclusion

The LAVE effect on many hematological RIs indicates the need for deliberate secondary exclusion for proper derivation of RIs. Obvious differences in Ghanaian RIs compared to other countries underscore the importance of country‐specific RIs for improved clinical decision‐making.

Establishing reference intervals for major biochemical analytes for the Russian population: a research conducted as a part of the IFCC global study on reference values

OBJECTIVES

Because reference intervals (RIs) for biochemistry analytes matched to the Russian population are not well defined, we joined the global study on reference values (RVs) coordinated by the IFCC Committee on Reference Intervals and Decision Limits (C-RIDL).

METHODS

According to the C-RIDL harmonized protocol, 793 healthy volunteers were recruited in Saint-Petersburg, Moscow, and Yekaterinburg. Serum samples were tested for 34 biochemistry analytes. Sources of variation of RVs were explored using multiple regression analysis. The need for partitioning RVs by sex and age were judged using standard deviation ratio based on ANOVA. Latent abnormal values exclusion (LAVE) method was applied to reduce the influence of individuals with metabolic syndrome and/or inappropriate sampling conditions. RIs were computed by the parametric method.

RESULTS

No appreciable between-city differences were observed. Partition of RVs by sex was required for 17 analytes. Age-related changes in RVs were observed in many analytes, especially in females. The trend was exaggerated in nutritional and inflammatory markers that were closely associated with body mass index (BMI), because BMI increases prominently with age. Therefore, for those analytes, volunteers with BMI > 28 kg/m² were excluded in determining RIs for age-specific RIs. The LAVE method was effective in lowering the upper limits of the RIs for nutritional and inflammatory markers.

CONCLUSION

RIs matched to the Russian population were established for 34 biochemical analytes using up-to-date methods in detailed consideration of sources of variation of RVs. The majority of Russian RIs are similar to those of Caucasian populations among the participating countries.

Establishment of reference intervals for immunoassay analytes of adult population in Saudi Arabia

Background

This is a second part of report on the IFCC global multicenter study conducted in Saudi Arabia to derive reference intervals (RIs) for 20 immunoassay analytes including five tumor makers, five reproductive, seven other hormones and three vitamins.

Methods

A total of 826 apparently healthy individuals aged ≥18 years were recruited in three clinical laboratories located in western, central and eastern Saudi Arabia using the protocol specified for the global study. All serum specimens were measured using Abbott, Architect analyzers. Multiple regression analysis (MRA) was performed to explore sources of variation of each analyte: age, body mass index (BMI), physical exercise and smoking. The magnitude of variation of reference values (RVs) attributable to sex, age and region was calculated by ANOVA as a standard deviation ratio (SDR). RIs were derived by the parametric (P) method.

Results

MRA revealed that region, smoking and exercise were not relevant sources of variation for any analyte. Based on SDR and actual between-sex differences in upper limits (ULs), we chose to partition RIs by sex for all analytes except for α-fetoprotein and parathyroid hormone (PTH). Age-specific RIs were required in females for ferritin, estradiol, progesterone, testosterone, follitropin, luteotropin and prolactin (PRL). With prominent BMI-related increase, RIs for insulin and C-peptide were derived after excluding individuals with BMI > 32 kg/m2. Individuals taking vitamin D supplements were excluded in deriving RIs for vitamin D and PTH.

Conclusions

RIs of major immunoassay analytes specific for Saudi Arabians were established in careful consideration of various biological sources of variation.