Testing the limits: the diagnostic accuracy of reference change values.

Reference change values (RCVs) are used by the physician to judge whether a change in analyte concentration from one sample to the next may represent a clinically significant change. Published RCVs are usually given as fixed percentages of the analyte concentration in the first sample. The accuracy of published RCVs is not well known. We obtained public-use data from the US National Health and Nutrition Examination Survey (NHANES) 2001-2002 to study the distribution of changes in the concentration of eight commonly used analytes. Specimens were obtained on two occasions 7-47 days apart from 279 to 411 individuals with an analyte concentration within the reference interval in both samples. The analytes were albumin, calcium, cholesterol, phosphate, potassium, sodium, hemoglobin and thrombocytes. For each analyte, normal within-subject biological coefficient of variation from the EFLM Working Group on Biological Variation and the NHANES analytical coefficient of variation were used to calculate the 5 and 95 percentile RCVs. These RCVs were calculated as fixed percentages of the analyte concentrations in the first sample and compared to the empirical 5 and 95 percentiles. The sensitivity of the RCVs in detecting changes outside the empirical percentiles ranged from 0.35 for sodium to 0.80 for albumin. The specificity of the RCVs in detecting changes inside the empirical percentiles ranged from 0.85 for potassium to 0.97 for thrombocytes. Calculating RCVs as fixed percentages of the analyte concentration in the first sample lessened the diagnostic accuracy. RCVs given as a function of the first result would perform better.

Sustained Maternal Hyperandrogenism During PCOS Pregnancy Reduced by Metformin in Non-obese Women Carrying a Male Fetus.

CONTEXT: Large, longitudinal studies on androgen levels in pregnant women with polycystic ovary syndrome (PCOS) are lacking. While metformin has a mild androgen-lowering effect in non-pregnant women with PCOS, its effects on maternal androgen levels in pregnancy are less well understood. OBJECTIVE: To describe androgen patterns in pregnant women with PCOS and in healthy control women, and to explore the potential effects of metformin on maternal androgen levels in PCOS. DESIGN AND SETTING: A post hoc analysis from a randomized, placebo-controlled, multicenter study carried out at 11 secondary care centers and a longitudinal single-center study on healthy pregnant women in Norway. PARTICIPANTS: A total of 262 women with PCOS and 119 controls. INTERVENTION: The participants with PCOS were randomly assigned to metformin (2 g daily) or placebo, from first trimester to delivery. MAIN OUTCOME MEASURES: Androstenedione (A4), testosterone (T), sex-hormone binding globulin (SHBG), and free testosterone index (FTI) at 4 time points in pregnancy. RESULTS: Women with PCOS versus healthy controls had higher A4, T, and FTI, and lower SHBG at all measured time points in pregnancy. In the overall cohort of women with PCOS, metformin had no effect on A4, T, SHBG, and FTI. In subgroup analyses, metformin reduced A4 (P = 0.019) in nonobese women. Metformin also reduced A4 (P = 0.036), T (P = 0.023), and SHBG (P = 0.010) levels through pregnancy in mothers with a male fetus. CONCLUSION: Metformin had no effect on maternal androgens in PCOS pregnancies. In subgroup analyses, a modest androgen-lowering effect was observed in nonobese women with PCOS. In PCOS women carrying a male fetus, metformin exhibited an androgen-lowering effect.

Establishing the 99th percentile of a novel assay for high-sensitivity troponin I in a healthy blood donor population.

Background: The recommended cut-off of cardiac troponin (cTn) for the diagnosis of acute myocardial infarction (AMI) is the 99th percentile in a healthy reference population. We aimed to determine the 99th percentile of the novel ADVIA Centaur® High Sensitivity Troponin I assay (Siemens Healthcare Diagnostics) in fresh lithium heparin plasma samples from healthy blood donors. Methods: A total of 1000 apparently healthy blood donors were included. High-sensitivity (hs) cTnI, hs-cTnT, creatinine and N-terminal pro b-type natriuretic peptide (NT-proBNP) were measured in fresh lithium heparin plasma samples, and glycated hemoglobin (HbA1c) was measured in ethylenediaminetetraacetic acid (EDTA)-blood. The 99th percentile was estimated for the whole population, as well as for males and females separately. Results: For the total population the 99th percentile of ADVIA Centaur® High Sensitivity Troponin I was 96 (65-149) ng/L. The estimated value differed significantly from results published by others and was highly dependent on which values were considered statistical outliers. Conclusions: The estimated 99th percentile for hs-cTnI in the population studied differed significantly from previously published results. There is a need for further specifications regarding how subjects used for estimating the 99th percentile of cTns in healthy populations should be recruited and how outlier values should be identified, as this can highly influence the diagnostic cut-off applied for AMI.

A new index of clinical utility for diagnostic tests.

Clinical utility of a diagnostic test depends on its diagnostic accuracy, the pretest probability of disease and the clinical consequences of the test results. Tools for evaluating clinical utility are scarce. We propose a new clinical utility index (CUI), which is the expected gain in utility (EGU) of the test divided by the EGU of an ideal test, both adjusted for EGU of the optimal clinical action without testing. The index expresses the relative benefit of using the test compared to using an optimal test when making a clinical decision. To illustrate how the index may be used, we estimated CUI for fasting glucose, both as a continuous and as a dichotomous test, at several values of pretest probability of diabetes mellitus and at two levels of cost/benefit-ratio. In the same clinical situations we also estimated CUI for the 2 h glucose tolerance test. Hemoglobin A1c ≥ 48 mmol/mol was used as a reference standard for diabetes mellitus. In this model, fasting glucose was clinically more useful as a continuous test than as a dichotomous one, based on CUIs. At pretest probability above the treatment threshold, fasting glucose as a continuous test was even more useful than the complete glucose tolerance test. These results are not necessarily generalizable; however, they show how the CUI can be used to select the most useful test in certain clinical situations.

Allowable bias derived from the NOBIDA reference values.

Figures of allowable bias are used to make rational choices of quality control rules, to judge the validity of published reference values, and to determine the stability of sample materials. Usually, allowable bias is parametrically defined as 0.25 times the total biological standard deviation, because that is half the width of the 90% confidence interval of parametrically estimated reference limits from 120 reference values. The published figures are mostly derived from very small populations, less than 120. We estimated allowable bias non-parametrically as the least of 4 percentile differences in distributions of reference values from the large Nordic reference interval project biobank and database (NOBIDA). The percentile differences are equivalent to 0.25 times the total biological standard deviation in Gaussian distributions. We also estimated allowable bias from the distributions of non-parametrically estimated reference limits after resampling 120 reference values from the same datasets. Clearly larger allowable bias was derived from the resampling method than from the percentile difference method, showing that non-parametric estimation of reference limits from 120 reference values implies a larger allowable bias than 0.25 times the normal biological standard deviation. With some exceptions, the figures of allowable bias using the percentile difference method were in the same order of magnitude as parametrically derived figures in other studies, and lend some support to the results from those smaller studies. Whether such bias specifications, if met, guarantee measurements of sufficient clinical quality is unknown.

Hemoglobin A1c as screening for gestational diabetes mellitus in Nordic Caucasian women.

BACKGROUND: Gestational diabetes mellitus (GDM) increases the risk for preeclampsia and macrosomia. GDM is conventionally diagnosed by an oral glucose tolerance test (OGTT). Hemoglobin A1c (HbA1c) is a marker for the average glucose level the last 2-3 months. We aimed to study if HbA1c alone or in combination with patient characteristics can be used to screen for GDM and reduce the number of OGTTs, and whether it could predict preeclampsia or birth weight. METHODS: 855 women from a previous study on the effect of exercise on GDM prevalence were eligible, whereof 677 were included. GDM was diagnosed by WHO 1999 criteria (GDM-WHO) and modified IADPSG criteria (GDM-IADPSG), at pregnancy weeks 18-22 and 32-36. HbA1c analyzed at pregnancy weeks 18-22 and 32-36, variables from patient history and clinical examination were considered for logistic regression models. The diagnostic accuracy was assessed by ROC curve analysis. RESULTS: Accumulated GDM prevalence was 6.7 % by WHO and 7.2 % by modified IADPSG criteria. Nearly a third could potentially have avoided an OGTT by using HbA1c to exclude GDM-IADPSG with a sensitivity of 88 % at week 18-22 and 97 % at week 32-36. Further, 16 % could have avoided an OGTT with a sensitivity of 96 % using HbA1c at week 18-22 to exclude GDM-IADPSG throughout pregnancy. HbA1c was not accurate at diagnosing GDM-IADPSG, and it was inaccurate at screening for GDM-WHO at any time point. Adding other predictors did not increase the number of potentially avoidable OGTTs significantly. HbA1c was not significantly associated with preeclampsia or birth weight. CONCLUSIONS: HbA1c could potentially reduce the number of OGTTs.

Using the hazard ratio to evaluate allowable total error in predictive measurands.

BACKGROUND: Allowable total error is usually derived from data on biological variation or from state-of-the-art of measuring technology. Here we present a new principle for evaluating allowable total error when the concentration of the analyte (the measurand) is used for prediction: What are the predictive consequences of allowable total errors in terms of errors in the estimate of the hazard ratio (HR)? METHODS: We explored the effect of analytical measurement errors on Cox regression estimates of HR. Published data on Cox regression coefficients were used to illustrate the effect of measurement errors on predicting cardiovascular events or death based on serum concentration of cholesterol and on progression of chronic kidney disease to kidney failure based on serum concentrations of albumin, bicarbonate, calcium and phosphate, and urine albumin/creatinine-ratio. RESULTS: If the acceptable error in the estimate of the HR is 10%, allowable total errors in serum cholesterol, bicarbonate and phosphate are approximately the same as allowable total error based on biological variation, while allowable total error in serum albumin and calcium are a little larger than estimates based on biological variation. CONCLUSIONS: Evaluating allowable total error from its effect on the estimate of HR is universally applicable to measurands used for prediction.

HbA1c as screening for gestational diabetes mellitus in women with polycystic ovary syndrome.

BACKGROUND: Gestational diabetes mellitus (GDM) is associated with adverse pregnancy outcomes such as preeclampsia and macrosomia. Women with polycystic ovary syndrome (PCOS) are at increased risk of developing GDM. Today, GDM is diagnosed by oral glucose tolerance test (OGTT), a rather cumbersome test for the women and health care system. The objectives of this study were to investigate whether HbA1c in first trimester of pregnancy could be used as a screening test for GDM in first trimester and throughout pregnancy in order to reduce the number of OGTTs, and whether it could predict preeclampsia and macrosomia in women with PCOS. METHODS: Post hoc analyses of data from 228 women from a prospective, randomised, multicenter study comparing metformin to placebo from first trimester to delivery. Fasting and 2-h plasma glucose were measured during a 75 g OGTT in first trimester, gestational week 19 and 32 as well as fasting plasma glucose in gestational week 36. GDM was diagnosed by WHO criteria from 1999 in first trimester and throughout pregnancy and by modified IADPSG criteria (i.e. lacking the 1-h plasma glucose value) in first trimester. The diagnostic accuracy was assessed by logistic regression and ROC curve analysis. RESULTS: The area under the ROC curve for first trimester HbA1c for screening of GDM diagnosed by WHO criteria in first trimester was 0.60 (95 % CI 0.44-0.75) and 0.56 (95 % CI 0.47-0.65) for GDM diagnosed throughout pregnancy. Only 2.2 % (95 % CI 0.7-5.1 %) of the participants could have avoided OGTT. HbA1c was not statistically significantly associated with GDM diagnosed by modified IADPSG criteria in first trimester. However, first trimester HbA1c was statistically significantly associated with preeclampsia. Both HbA1c and GDM by WHO criteria in first trimester, but not by IADPSG, were negatively associated with birth weight. CONCLUSION: First trimester HbA1c can not be used to exclude or predict GDM in women with PCOS, but it might be better to predict preeclampsia than the GDM diagnosis.

Using the likelihood ratio to evaluate allowable total error--an example with glycated hemoglobin (HbA1c).

BACKGROUND: Allowable total error is derived in many ways, often from data on biological variation in normal individuals. We present a new principle for evaluating allowable total error: What are the diagnostic consequences of allowable total errors in terms of errors in likelihood ratio (LR)? Glycated hemoglobin A1c in blood (HbA1c) in diagnosing diabetes mellitus is used as an example. Allowable total error for HbA1c is 3.0% derived from data on biological variation compared to 6.0% as defined by National Glycohemoglobin Standardization Program (NGSP). METHODS: We estimated a function for LR of HbA1c in diagnosing diabetes mellitus using logistic regression with a clinical database (n=572) where diabetes status was defined by WHO criteria. Then we estimated errors in LR that correspond to errors in the measurement of HbA1c. RESULTS: Measuring HbA1c 3% too low at HbA1c of 6.5 percentage points (the suggested diagnostic limit) gives a LR of 0.36 times the correct LR, while measuring HbA1c 3% too high gives a LR of 2.77 times the correct LR. The corresponding errors in LR for allowable total error of 6% are 0.13 and 7.69 times the correct LR, respectively. CONCLUSIONS: These principles of evaluating allowable total error can be applied to any diagnostically used analyte where the distribution of the analyte's concentration is known in patients with and without the disease in a clinically relevant population. In the example used, the allowable total error of 6% leads to very erroneous LRs, suggesting that the NGSP limits of ±6% are too liberal.

Dry ice exposure of plasma samples influences pH and lupus anticoagulant analysis.

BACKGROUND: Tests for lupus anticoagulant (LA), including silica clotting time (SCT) and diluted Russel's viper venom time (dRVVT) are used to diagnose antiphospholipid syndrome. Due to sample instability, it is recommended that samples are frozen if analysis is postponed >4 h. Shipping on dry ice is common practice to keep samples frozen during transport. Recent data suggest that exposure to dry ice may affect sample pH and results in subsequent analyses. We aimed to determine the effect of dry ice on pH and LA analysis. METHODS: Citrated plasma from eight healthy volunteers was allocated to three preanalytical regimes: 1) storage at -20 °C; 2) dry ice exposure followed by storage at -20 °C; or 3) dry ice exposure followed by storage at -80 °C. RESULTS: Samples stored at -20 °C after dry ice exposure had significantly lower median pH (1.2 units, p=0.01) and prolonged clotting time ratios (up to 55% for dRVVT tests, p<0.02) in LA analysis, compared to samples not exposed to dry ice. This resulted in poor test specificity (25%). Similar changes were not observed in samples placed at -80 °C after dry ice exposure. CONCLUSIONS: Dry ice may affect sample pH and increase the fraction of false positive LA results. This preanalytical factor should be taken into account by laboratories receiving frozen samples for these tests.