Distinguishing reference intervals and clinical decision limits - A review by the IFCC Committee on Reference Intervals and Decision Limits

Reference Intervals (RIs) and clinical decision limits (CDLs) are a vital part of the information supplied by laboratories to support the interpretation of numerical clinical pathology results. RIs describe the typical distribution of results seen in a healthy reference population while CDLs are associated with a significantly higher risk of adverse clinical outcomes or are diagnostic for the presence of a specific disease. However, as the two concepts are sometimes confused, there is a need to clarify the differences between these terms and to ensure they are easily distinguished, especially because CDLs have a clinical association with specific diseases and risks, thereby implying that effective clinical interventions are available. It is important to note that, because population-based RIs are derived from the range of values expected in a typical community population, laboratory results that fall outside a RI do not necessarily indicate a disease but rather that additional medical follow-up and/or treatment may be warranted. In contrast, CDLs are associated with a risk of specific adverse outcomes, and are commonly used to interpret laboratory test results, including lipid parameters, glucose, hemoglobin A1c (HbA1c), and tumor markers, to determine risk of disease, to diagnose or to treat. In recent years, the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) Committee on Reference Intervals and Decision Limits (C-RIDL) has focused primarily on RIs and has performed multicenter studies to obtain common RIs. However, the broader responsibility of the Committee, from its name, includes "decision limits". C-RIDL now aims to emphasize the importance of the correct use of both RIs and CDLs and to encourage laboratories to specify the appropriate information to clinicians as needed. This review discusses RIs and CDLs in detail, describes the similarities and the differences between these two important tools in laboratory medicine, and clearly explains the processes used to define them. C-RIDL encourages the involvement of laboratory professionals in the establishment of both RIs and CDLs.

The Impact of Age on the Delay in Diagnosis in Patients With Acute Pulmonary Embolism

It has been speculated that the atypical clinical presentation of acute pulmonary embolism (PE) in older patients leads to a late diagnosis and therefore contributes to a worse prognosis. Therefore, we prospectively evaluated the delay in diagnosis and its relation to the in-hospital mortality in 202 patients with acute PE. Patients >65 years presented more often with hypoxia ( P = .017) and with a history of syncope ( P = .046). Delay in diagnosis was not statistically different in both age groups. Older age was significantly associated with an increased risk for in-hospital mortality (OR 4.36, 95% CI 0.93-20.37, P = .043), whereas the delay in diagnosis was not associated with an increase of in-hospital mortality. We therefore conclude that the clinical presentation of acute PE in older patients cannot be considered as a risk factor for late diagnosis and is not responsible for their higher in-hospital death rate.

Reference limits of plasma and serum creatinine concentrations from intra-laboratory data bases of several German and Italian medical centres: Comparison between direct and indirect procedures

BACKGROUND

The current dogma of establishing intra-laboratory reference limits (RLs) and their periodical reviewing cannot be fulfilled by most laboratories due to the expenses involved. Thus, most laboratories adopt external sources for their RLs often neglecting the problems of transferability. Therefore, several attempts were undertaken to derive RLs from the large data pools stored in modern laboratory information systems. These attempts were further developed to a more sophisticated indirect procedure. The new model can be considered a combined approach because it pre-excludes some subjects by direct criteria. In the current study, the new concept was applied to estimate RLs for serum and plasma creatinine from several German and Italian laboratories.

METHODS

A smoothed kernel density function was estimated for the distribution of the total mixed data of the sample group (combined data of non-diseased and diseased subjects). It was assumed that the "central" part of the distribution of all data represents the non-diseased ("healthy") population. The central part was defined by truncation points using an optimisation method, and was used to estimate a Gaussian distribution of the values of presumably non-diseased subjects after Box-Cox transformation of the empirical data. This distribution was now considered as the distribution of the non-diseased subgroup. The percentiles of this parametrical distribution were calculated to obtain RLs.

RESULTS

RLs determined by the indirect combined decomposition technique led to similar RLs as the classical direct method. Furthermore, the RLs obtained from 14 laboratories in 2 different European regions reflected the well-known differences of various analytical procedures. Stratification for gender and age was necessary. With rising age, an increase of the upper RL and of the reference range was observed. Hospitalization appeared also to affect the RLs. The new approach led to RLs in an artificially mixed population of diseased and non-diseased subjects (selected by clinical criteria) which were identical to RLs determined by a direct method applied to the non-diseased subgroup.

CONCLUSIONS

The proposed strategy of combining exclusion criteria with a resolution technique led to plausible retrospective RLs from intra-laboratory data pools for creatinine. Differences between laboratories were mainly due to the well-known bias of the different analytical procedures.

Reference intervals for serum creatinine concentrations: assessment of available data for global application

BACKGROUND

Reference intervals for serum creatinine remain relevant despite the current emphasis on the use of the estimated glomerular filtration rate for assessing renal function. Many studies on creatinine reference values have been published in the last 20 years. Using criteria derived from published IFCC documents, we sought to identify universally applicable reference intervals for creatinine via a systematic review of the literature.

METHODS

Studies were selected for inclusion in the systematic review only if the following criteria were met: (a) reference individuals were selected using an "a priori" selection scheme, (b) preanalytical conditions were adequately described; (c) traceability of the produced results to the isotope dilution-mass spectrometry (IDMS) reference method was demonstrated experimentally, and (d) the collected data received adequate statistical treatment.

RESULTS

Of 37 reports dealing specifically with serum creatinine reference values, only 1 report with pediatric data and 5 reports with adult data met these criteria. The primary reason for exclusion of most papers was an inadequate demonstration of measurement traceability. Based on the data of the selected studies, we have collated recommended reference intervals for white adults and children.

CONCLUSION

Laboratories using methods producing traceable results to IDMS can apply the selected reference intervals for serum creatinine in evaluating white individuals.

Reference intervals for the glomerular filtration rate and cell-proliferation markers: serum cystatin C and serum beta 2-microglobulin/cystatin C-ratio

Recent studies have indicated that serum and plasma cystatin C are better markers for glomerular filtration rate (GFR) than serum creatinine, ubiquitously used for this purpose. To fully exploit the value of serum and plasma cystatin C as GFR markers, reliable age and sex-correlated reference intervals are required. The present study comprised cystatin C determinations in plasma and sera from 259 individuals from a well-defined area in the southernmost part of Sweden. From demographic lists two men and two women were randomly selected from each one-year birth cohort above 20 years of age. No sex differences were found for plasma and serum cystatin C, whereas an increase in the cystatin C levels with age was noted, corresponding to the known age-related decrease in GFR. The following reference intervals are recommended for practical clinical use: S-Cystatin C (both sexes): 20-50 years, 0.70-1.21 mg l-1 and 50+ years, 0.84-1.55 mg l-1. The same samples were also used for determination of beta 2-microglobulin levels in order to calculate reference intervals for the beta 2-microglobulin/cystatin C-ratio, which is a more distinct marker for cell proliferation, particularly lymphoproliferation, than is the serum level of beta 2-microglobulin alone, since the ratio should be virtually uninfluenced by GFR. The beta 2-microglobulin/cystatin C-ratios were uninfluenced by sex and age and 1.45-2.43 is recommended as the serum reference interval for practical clinical use. Serum creatinine was determined in the same samples and the creatinine level was found to be strongly influenced by sex and weakly by age.

Serum uric acid and related factors in 500 hospitalized subjects

The study purpose was to determine the following in a large sample of hospitalized patients: (1) the prevalence of hyperuricemia, (2) the association of hyperuricemia with other metabolic disorders, and (3) the factors independently predicting hyperuricemia. Five hundred adult patients (250 men and 250 women) were randomly selected from those admitted as inpatients over a period of 5 months. In all patients, body mass index (BMI), blood pressure, and serum glucose, lipid, creatinine, urea nitrogen, and urate concentrations were measured. The presence of diseases or use of medications known to affect serum urate levels were recorded. The mean level of serum urate was 5.6 mg/dL in the whole sample, 6.0 mg/dL in men and 5.3 mg/dL in women (P = .003, men v women). The prevalence of hyperuricemia was 27.6% (28.8% and 26.4% in men v women, P = nonsignificant). A definite or probable secondary hyperuricemia was found in 87.7% of the subjects. Hyperuricemia was rarely isolated (21%), whereas it was frequently associated with hypertension (60.1%), hyperlipidemia (31.2%), diabetes (28.3%), and obesity (21.7%). In 26.8% of the subjects, hyperuricemia was associated with two metabolic disorders, in 13.8% with three, and in 2.9% with four. Multiple metabolic disorders (three to four) were found in 16.7% of subjects with hyperuricemia. Serum urate levels progressively increased across a range of subjects from those without diabetes, hyperlipidemia, hypertension, or obesity to those with one, two, or a greater number of associated metabolic abnormalities. Multiple stepwise regression analysis showed that 43% of serum urate variability was explained by urea nitrogen levels, triglyceride levels, diuretic therapy, the inverse of creatinine (as an index linearly related to creatinine clearance), and BMI. These results indicate that in hospitalized subjects, hyperuricemia is (1) frequent, (2) a secondary phenomenon in most cases, and (3) frequently associated with other metabolic disorders. The major predictors of high serum urate levels are BMI, triglycerides, parameters of renal function, and use of diuretics. These variables explain a large proportion of serum urate variability.

Diagnosis of mixed acid-base disorders in diabetic ketoacidosis

In diabetic ketoacidosis, a mixed acid-base disorder is suggested when the anion gap increase (delta AG) does not equal the bicarbonate decrease (delta HCO3), or when the delta AG/delta HCO3 ratio does not equal 1.0. It is widely assumed that delta AG/delta HCO3 is significantly different from 1.0 when it is less than 0.8 or greater than 1.2. The validity of these ratio limits were examined by analyzing a normal control group of 68 subjects and 27 diabetic ketoacidosis admissions that had no evidence of mixed disorders. In the 27 ketoacidosis admissions, regression analysis showed that delta AG was predicted to equal delta HCO3, as expected in pure anion gap acidosis: delta AG = 1.0 delta HCO3 (r = 0.744, p < 0.001). It was found that delta AG is significantly different from delta HCO3 when they differ by more than 8 mEq/L, and equivalently, delta AG/delta HCO3 is significantly different from 1.0 when it is less than (1.0 - 8/delta HCO3) or greater than (1.0 + 8/delta HCO3). These criteria from regression analysis suggested that 4% of the 27 pure anion gap acidoses, and 3% of the control group, had mixed disorders. In contrast, the ratio limits of 0.8 and 1.2 suggested 56% of the pure anion gap acidoses, and 94% of the control group, had mixed disorders. It was concluded that mixed disorders are overdiagnosed by the ratio limits of 0.8 and 1.2. Mixed disorders are more accurately detected by noting whether delta AG and delta HCO3 differ by more than 8 mEq/L.

Factors influencing haematological measurements in healthy adults

By studying 516 healthy adults normal reference intervals were established for the Coulter "S" haematological indices with the plasma ferritin, B12, folate and red cell folate in a subgroup of 306. Significant sex related differences were found for all measurements other than MCV, MCH and B12. After allowing for these sex related differences, the effects of age, body size, fasting, smoking, alcohol, exercise and contraceptive pill usage on the parameters studied was defined.

Anion gap-bicarbonate relation in diabetic ketoacidosis

The relation between the serum anion gap and the serum total carbon dioxide concentration was studied in 100 admissions of patients with diabetic ketoacidosis and 43 normal control subjects. In 20 admissions of patients with diabetic ketoacidosis (Group 1), the patients had no other conditions or medications known to alter acid-base or electrolyte homeostasis, whereas in 80 admissions of patients with diabetic ketoacidosis (Group 2), the patients had at least one of these factors. Analysis of the change in total carbon dioxide compared with the change in anion gap in Group 1 and control subjects revealed the following relation: change in total carbon dioxide = 0.74 + 1.00 X change in anion gap, in meq/liter (r = 0.886, p less than 10(-7]. The 95 percent prediction interval for detecting mixed acid-base disorders with this equation was +/- 8 meq/liter. Analysis of all admissions of patients with diabetic ketoacidosis and control subjects combined showed that the anion gap increased 0.24 meq/liter per mg/dl increase in blood urea nitrogen (with total carbon dioxide constant). Because the highest blood urea nitrogen level in Group 1 and control subjects was 22 mg/dl, the change in total carbon dioxide-change in anion gap regression is generally not valid for blood urea nitrogen levels higher than 22 mg/dl. Thus, both the wide prediction interval and volume depletion (as reflected by blood urea nitrogen level) impair the usefulness of the anion gap as a screen for mixed acid-base disorders in patients with diabetic ketoacidosis.