Computation of profile likelihood-based confidence intervals for reference limits with covariates.

Many biochemical quantities depend on age or some other covariate. Reference limits that allow for these dependencies help physicians to interpret the results of biochemical tests. Because reference limits must be estimated, it is important to assess their precision with, for example, confidence intervals. This paper relies on the assumption that data can be modeled by a generalized linear model and presents a method for calculating approximate profile likelihood-based confidence intervals for reference limits. The calculation of confidence intervals is based on a new method that draws on profile likelihood-based confidence intervals in general statistical models. The asset of this new method is that only two constrained optimization problems have to be solved instead of several in the standard method. We motivate our confidence interval calculation method with two applications. The first is for data on immunoglobulin concentration in the context of a generalized linear model with gamma distribution. This model is compared with the often used lognormal model. The second application handles data on serum alpha-fetoprotein and is presented in a linear regression situation. In the latter application the widths of the calculated profile confidence intervals are compared with exact and approximate regression-based intervals and the actual confidence levels are determined by simulation.

Regression-based reference limits: determination of sufficient sample size.

Regression analysis is the method of choice for the production of covariate-dependent reference limits. There are currently no recommendations on what sample size should be used when regression-based reference limits and confidence intervals are calculated. In this study we used Monte Carlo simulation to study a reference sample group of 374 age-dependent hemoglobin values. From this sample, 5000 random subsamples, with replacement, were constructed with 10-220 observations per sample. Regression analysis was used to estimate age-dependent 95% reference intervals for hemoglobin concentrations and erythrocyte counts. The maximum difference between mean values of the root mean square error and original values for hemoglobin was 0.05 g/L when the sample size was > or = 60. The parameter estimators and width of reference intervals changed negligibly from the values calculated from the original sample regardless of what sample size was used. SDs and CVs for these factors changed rapidly up to a sample size of 30; after that changes were smaller. The largest and smallest absolute differences in root mean square error and width of reference interval between sample values and values calculated from the original sample were also evaluated. As expected, differences were largest in small sample sizes, and as sample size increased differences decreased. To obtain appropriate reference limits and confidence intervals, we propose the following scheme: (a) check whether the assumptions of regression analysis can be fulfilled with/without transformation of data; (b) check that the value of v, which describes how the covariate value is situated in relation to both the mean value and the spread of the covariate values, does not exceed 0.1 at minimum and maximum covariate positions; and (c) if steps 1 and 2 can be accepted, the reference limits with confidence intervals can be produced by regression analysis, and the minimum acceptable sample size will be approximately 70.

Regression-based reference limits and their reliability: example on hemoglobin during the first year of life.

Calculation of reference limits by regression analysis makes it unnecessary to partition the reference data into subgroups, and age-dependent limits can be estimated within as narrow age intervals as necessary. However, the reliability of the regression-based reference limits has not been considered before. To get valid regression-based confidence intervals (CIs) for reference limits, one must evaluate the convolution of two distributions. In this study, age-dependent reference limits with corresponding CIs were produced for blood hemoglobin concentrations over the age interval from newborns to 12 months. We describe how the variance associated with the reference limits can be estimated, and present a Table from which appropriate values can be chosen for the calculation of regression-based reference limits and exact CIs. Also, an equation for the calculation of approximate CIs is given. The data were modeled by linear regression in several cumulative age groups to find the transition zone where the slope changed. After defining this cutoff point, piecewise linear regression was applied. Reference limits and their CIs calculated by conventional and piecewise linear regression methods were almost the same in older age groups but differed significantly during the period of most rapid age-dependent changes, i.e., during the 2 months after birth.

Perceptual magnet effect in the light of behavioral and psychophysiological data.

Finnish speaking adults categorized synthetic vowels, varying in the frequency of the second formant (F2), as either /y/ or /i/. Two subject groups emerged: "good" and "poor" categorizers. In a /i/ rating experiment, only the good categorizers could consistently label their best /i/ (the prototype, P), being low in the F2 continuum. Poor categorizers rated /i/'s with high F2 values as Ps. In a same/different (AX) discrimination experiment, using the individual Ps and nonprototypes (NPs), it was more difficult for good categorizers to detect small F2 deviations from the P than from an NP (the "perceptual magnet effect"). For poor categorizers, the opposite effect was found. The same stimuli were used to record the mismatch negativity (MMN), an ERP component reflecting preattentive detection of deviations from a standard sound. For the good categorizers the MMNs were lower for Ps than for NPs; for the poor categorizers the MMNs for Ps and NPs did not differ significantly. The results show that individual listeners behaved differently in categorization and goodness rating but in the same way in attentive (AX) discrimination, being the poorest at about the same F2 location. The perceptual magnet effect was indicated in the good categorizers both by behavioral and psychophysiological (MMN) discrimination data.

Subchondral bone and cartilage repair with bioactive glasses, hydroxyapatite, and hydroxyapatite-glass composite.

The repair of an osteochondral defect in rabbit femur was studied with three kinds of bioactive glasses (BG), hydroxyapatite (HA), and hydroxyapatite-glass (HAG) composite. Seventy-two osteochondral defects were created in 18 rabbits. Sixty-four cylinders were implanted and eight defects were left empty as controls. Histomorphometry, scanning electron microscopy (SEM) and energy dispersive X-ray analysis (EDXA) were used for evaluation. Small osteochondral defects in rabbit femur found to heal themselves by regeneration. The three BGs, HA, and HAG led to direct lamellar bone repair of subchondral bone and restoration of articular surfaces mostly with hyalinelike cartilage in 12 weeks. However, the composition of the materials affects their behavior. Chondrogenesis took place earlier with the BGs than with HA. HAG degraded too much, glass 14 was too reactive and brittle, and the high alumina content in glass 11 disturbed its bone-bonding ability. Glass 7 and HA were the most balanced in the repair process. A special preparation method was used to retain soft tissues fairly unchanged and enable them to the observed together with hard tissues in SEM analysis.

Estimation of reference change limits using patient data.

Two approaches for deriving reference change limits from patient data are described. In the direct method, hospital database information is used for the selection of appropriate reference groups. If database information is not sufficient or reliable enough, but still most of the source data can be considered as health-related, an indirect method can be applied in the calculation of rough estimates for reference change limits. A computer program developed by us, GraphROC for Windows, includes both methods for the estimation of change limits from patient data. Time between specimen collections should be included as one classifying factor in the selection of source data. When only one previous result is available for comparison, change limits based on the reference sample group form the only available guide for clinical interpretation. However, when several previous results are available and the within-subject variances for the considered analyte are known to be heterogeneous between individuals, the clinical interpretation should rather be based on application of time series analysis.

Reliability and adequacy of discharge diagnosis databases in the production of reference values.

Discharge diagnoses provide a possibility to select patients individually and then to establish reference values for both "pathological" and control groups. Currently, the available diagnostic information is still at its infancy and should be carefully evaluated before the reference values based on those groups are utilized. It is anticipated that electronic storage of diagnostic and therapeutic information will be applied more commonly in the future as the development of computers makes it easier. The advanced utilization of laboratory data challenges physicians both in the clinical and laboratory side to participate in this development in order to make the information systems serve their actual needs more closely.

Reference intervals developed from data for hospitalized patients: computerized method based on combination of laboratory and diagnostic data.

We utilized the databases of a hospital information system to select for determination of reference values various individual hospitalized patients on the basis of their diagnoses at discharge. The nonparametric 2.5-97.5% "health-related" reference intervals were calculated for hemoglobin concentration, mean corpuscular volume (MCV), and erythrocyte count for both sexes. After excluding patients with diseases possibly affecting erythrocyte variables, we obtained a final group of 1786 women and 1450 men, ages 20-65 years, who were studied in age groups of 20-30, 30-45, 45-55, and 55-65 years. The upper reference limits of the MCV results obtained from hospitalized patients were higher than those produced conventionally from healthy individuals, as would be intuitively suggested by clinical experience. This method, based on selection by diagnosis, could be applicable to various analytes measured in hospital laboratories, provided sufficient data are available as databases.

Automatic discrimination of phonetically relevant and irrelevant vowel parameters as reflected by mismatch negativity.

An auditory event-related brain potential called mismatch negativity (MMN) was measured to study the perception of vowel pitch and formant frequency. In the MMN paradigm, deviant vowels differed from the standards either in F0 or F2 with equal relative steps. Pure tones of corresponding frequencies were used as control stimuli. The results indicate that the changes in F0 or F2 of vowels significantly affected the MMN amplitudes. The only variable significantly affecting the MMN latencies was sex which, however, did not have any effect on the amplitudes of the MMN. As expected, the MMN amplitudes increased with an increase in the acoustical difference between the standards and the deviants in all cases. On the average, the amplitudes were lower for the vowels than for the pure tones of equal loudness. However, in vowels, minor frequency changes in F0 produced higher MMN amplitudes than similar relative changes in F2. It was also noted that even the smallest and phonetically irrelevant change in F2 was detected by the MMN process. In overall, the results demonstrate that the MMN can be measured separately for F0 and F2 of vowels, although the MMN responses show large interindividual differences.

Method for determining reference changes from patients' serial data: example of cardiac enzymes.

Changes between serial laboratory test results can be significant, even if none of the individual results exceeds the reference interval. We developed a statistical method for the calculation of reference change limits from routine patients' data for situations in which the majority of the patients can be considered suitable reference subjects. The method was applied to cardiac enzyme data [creatine kinase (CK; EC 2.7.3.2), creatine kinase isoenzyme MB (CK-2), lactate dehydrogenase (LD; EC 1.1.1.28), and lactate dehydrogenase isoenzyme 1 (LD-1)] from 2029 consecutive patients. We used hospital discharge diagnoses to exclude patients with the diagnosis of myocardial infarction or myocarditis, but we also studied the characteristics of the method on unselected patients' data. The reference change limits derived from the diagnosis-selected patient group were as follows (U/L, activity measurements in serum at 37 degrees C according to Scandinavian recommendations): CK from -39 to 27, CK-2 from -8 to 7, LD from -86 to 85, and LD-1 from -19 to 15. Similar limits were obtained by conventional statistical methods from a group of 29 hospitalized patients with no myocardial symptoms. Our results suggest that it is possible to produce clinically applicable reference change limits from routine data.

On the nature of naming difficulties in aphasia.

An extensive naming battery was administered to ten patients representing classical aphasia syndromes. The battery included traditional performance measures and error scoring, phonological cuing, multiple-choice tasks tapping semantic and phonological knowledge, and word repetition tests. Differences in the patients' performance profiles were interpreted as reflecting lexical-phonological, phoneme assembly or multiple deficits. The results suggest that the hypothesized naming deficits have complex relationships to classical aphasia syndromes.

Quality control of multichannel hematology analyzers. Bivariate on-line comparison of MCV and MCH values for the detection of random errors.

Volume and hemoglobin content of red blood cells are physiologically interdependent variables but are measured independently in multichannel hematologic analyzers. Therefore correlation of these results to each other can be used as an additional step in internal quality control. To facilitate precise correlation, data were collected from 44,873 consecutive routine determinations of red blood cell parameters. Gaussian distributions were fitted to the original distributions and bivariate elliptical acceptance limits were calculated for simultaneous evaluation of mean corpuscular volume and mean corpuscular hemoglobin results. With the proposed method, rare combinations of the three directly measured red blood cell parameters (hemoglobin, red blood cell count, and mean corpuscular volume) can be identified for the detection of random analytical or preanalytical errors. This mean corpuscular volume-mean corpuscular hemoglobin rule could be added to internal quality-control programs of hematologic analyzers to improve detection of unusual results and possible random errors.

Measurement of sequential changes in plasma volume immediately after aortocoronary bypass surgery.

This study assessed the reliability of 131I albumin-dilution for measuring induced changes in plasma volume (PV) immediately after coronary bypass surgery. After injection of the tracer, blood samples were collected from 18 control patients to obtain 131I albumin decay curves in the plasma. These curves were used to construct a mathematical model describing tracer decay. PV was then calculated in control patients, and in 51 study patients who received plasma expanders on the first postoperative morning. At 1, 15, 30, and 60 min after volume loading in the study group, dilution volumes were calculated as the difference between predicted and observed plasma radioactivity levels. These calculations were compared to an independent measurement of PV. This method was associated with a measurement error no greater than +/- 12% in 82% of the 51 study patients. It tended to underestimate PV by 2.4%.