Narrow Confidence Limits Research Articles

Statistical issues of quality control in organised breast cancer screening.

European guidelines for breast-cancer screening recommend an integrated approach of mammography screening with subsequent assessment and biopsy, if required, in one screening unit under permanent quality control, for which target values are released. Although the calculation of the respective rates (e.g. for participation, assessment, biopsy, or cancer detection) appears trivial, the statistical assessment of their compatibility with the target values is less obvious. This is especially true if subjects with a positive diagnostic result leave the screening-assessment chain prematurely, and information about further diagnostic results outside the organised screening is lacking. Statistical models for the basic situation, in which complete information about the screening and assessment outcome is available, as well as for when information is incomplete, are presented. The statistical methods for obtaining the confidence limits, statistical tests and sample sizes needed to obtain a desired power of tests for the process parameters of interest are also given. The sample-size calculations indicate that large numbers of enrolled subjects are required to obtain reasonably narrow confidence limits, and that incomplete information about the outcome of diagnostic procedures among screening positives considerably worsens the feasibility of quality control. Although the methodology is specified for breast-cancer screening, it should be adaptable easily to other screening issues.

Confidence Limits for Intraclass Reliability Coefficients

A reliability coefficient (R) calculated using sample data may differ considerably in value from the population reliability coefficient. The reporting of confidence limits for the population reliability coefficient assists a person in interpreting the goodness of a sample reliability coefficient. The study was designed to (a) present the procedures for calculating confidence limits for R, (b) present confidence limits for several values of R and several sample sizes, and (c) describe a program in SPSS (1999) that provides confidence limits and hypothesis testing for Rs. The confidence limits for 3 different values of R, 2 different ANOVA models, 2 different sample sizes, 2 different numbers of repeated measures, and 2 different criterion scores were calculated. The confidence limits became narrower as R increased in value, as sample size increased in value, and as the number of repeated measures increased. For a given (a) R, (b) sample size, (c) number of repeated measures, and (d) criterion score, the confidence limits are the same for a 1-way and a 2-way Analysis of Variance model. Furthermore, when all other factors are equal, the confidence limits tend to be narrower for a criterion score that is a single score rather than a mean score. Based on the confidence limits reported, a sample size of at least 50 is needed to have relatively narrow confidence limits.

Improved Use of Inputs to Primordial Nucleosynthesis

We present a new method to incorporate laboratory nuclear data directly into the standard big-bang nucleosynthesis calculation. Using the quoted uncertainties on the data in a Monte Carlo procedure, we estimate likelihood distributions of the predicted abundances. Our results indicate significantly narrower confidence limits for the predictions than those presently in use. This technique provides error estimates that become smaller as more input data are added, and it allows new nuclear data to be incorporated easily as they become available.

SIPVADE: A New Computer Programme with Seventeen Statistical Tests for Outlier Detection in Evaluation of International Geochemical Reference Materials and its Application to Whin Sill Dolerite WS‐E from England and Soil‐5 from Peru

A new computer programme was written in programming language TURBOC, which enables us to apply a procedure involving seventeen statistical tests (a total of sixty five single or multiple outlier versions of these tests) for outlier detection in univariate sample at a high confidence level of 99% (significance level α= 0.01). The outlying observations should be evaluated first for technical reasons and then rejected manually from the data base until no more outliers are detected and the final statistical parameters are computed from the remaining data. This programme has been used successfully to process two reference material data bases: WS‐E from England and Soil‐5 from Peru. The final mean values for WS‐E are more reliable (characterized by smaller standard deviations and narrower confidence limits) than those obtained earlier using a different statistical approach. The application of a large number of statistical tests to Soil‐5 also resulted in smaller standard deviation values for most elements than the method involving a limited number of such tests. For WS‐E, some laboratories seem to have produced multiple data that were detected as statistical outliers. A close analysis of the distribution of outliers as a function of laboratory, country and analytical method leads to a technical justification for these outlying observations, probably in terms of inadequate QA/QC practices. Use of geochemical criteria indicates that the new mean values in WS‐E might be closer to the “true” concentrations. This procedure of outlier detection and elimination is therefore recommended in the study of the existing RM.

Stable carbon isotope ratios of latewood cellulose in Pinus sylvestris from northern Finland: variability and signal-strength

Stable carbon-isotope ratios were measured on the latewood cellulose of 25 Pinus sylvestris trees at five sites from the Arctic Circle to the northern limit of pine in Finland. A correlation-based analysis of variance was used to define within and between-site signal strengths. Expressed Population Signals (EPS) from samples of five trees are very high (0.89 to 0.95). Combining 15 trees from the three northernmost sites yields an EPS of 0.97. Despite the very strong common signal, different trees yield absolute δ13C values offset by >2%, which is similar in magnitude to the variability within individual series that includes the effect of climate. Samples of. 10 trees are necessary to obtain mean absolute values with confidence limits <1% wide (p < 0.05). The common practice of pooling four cores from four trees to yield absolute δ13C values is thus inadequate for palaeoclimate research. Because the δ13C values are homeoscedastic they can be indexed by differencing from the mean of each tree series, retaining all of the information on relative differences within each tree but yielding much narrower confidence limits when series are averaged. When five trees are used, 80% of annual confidence bands are, < ± 0.5%(P<0.05). Indexing effectively masks low-frequency variations when overlapping tree series are combined to produce long chronologies.

Normal Fetal Cardiac Dimensions Obtained by Perpendicular Imaging

This study provides normal fetal cardiac dimensional data in a large patient group over a wide range of gestational ages. Perpendicular imaging decreased lateral resolution error, resulting in normal values with narrower confidence limits than in prior studies.

Sixteen Statistical Tests for Outlier Detection and Rejection in Evaluation of International Geochemical Reference Materials: Example of Microgabbro PM‐S

A totally objective procedure involving sixteen statistical tests (a total of thirty four single or multiple outlier versions of these tests) for outlier detection and rejection in a univariate sample is applied to a data base of sixty four elements in a recently issued international geochemical reference material (RM), a microgabbro PM‐S from Scotland. This example illustrates the relative importance and usefulness of these tests in processing modern geochemical data for possible outliers and obtaining mean concentration and other statistical parameters from a final normal sample of univariate data. The final mean values are more reliable (characterized by smaller standard deviations and narrower confidence limits) than those obtained earlier using an accommodation approach (robust techniques) applied to this data base. Very high quality (certified value equivalent, cve) mean data are now obtained for eleven elements as well as high quality recommended values (rv) for thirty three elements in PM‐S. Earlier work using the accommodation approach failed to establish even one cve value for any of the sixty four elements compiled here. The present procedure of outlier detection and elimination is therefore recommended in the study of RMs

Human Immunodeficiency Virus type 1 infection and breast milk

Major questions are whether mothers infected with the Human Immunodeficiency Virus type 1 (HIV‐1) transmit the virus through breast milk and the magnitude of the additional transmission risk. The demonstration of a dose‐response effect is an epidemiological method to demonstrate causality. Thus, a study was carried out by the Italian Register for HIV Infection in Children on 961 children of known infection status. Duration of breast‐feeding was considered as the level of exposure in 168 ever breastfed children. Results showed that duration of practice significantly increased the risk of transmission. The adjusted infection odds ratio for one day of breast‐ versus exclusive formula‐feeding was 1.19 with narrow confidence limits (1.10–1.28). In a second study by the Register on 556 children of known infection status and derived prospectively, an infection odds ratio of 2.55 (confidence interval: 1.03–6.37) was calculated in breast‐ versus exclusively formula‐fed children. Several lines of evidence, including the above‐mentioned data from the Italian Register for HIV Infection in Children, showed a contribution of breast‐feeding to mother‐to‐child HIV‐1 transmission. Thus, this practice is now discouraged in HIV‐1 infected mothers living in industrialized societies where formula feeding is practical and attainable. Mode of feeding was known in 2183 children enrolled in the Register and born to HIV‐1 infected mothers since 1981. It could be observed that feeding habits of at‐risk infants changed in Italy in the middle 1980s, when a large majority of subjects was identified at birth. However, women infected exclusively by the sexual route are often unconscious of being infected or even being at risk from infection and consequently their children are less frequently identified at birth than are children of women with a history of intravenous drug use. In industrialized areas, the former children remain at risk from milk‐borne HIV‐1 infection.

A Noninvasive Strategy for the Treatment of Patients With Suspected Pulmonary Embolism

Pulmonary embolism has historically presented a formidable diagnostic problem because of the nonspecificity of the clinical findings associated with this disorder and the diagnostic uncertainties and challenges presented by both ventilation-perfusion lung scanning and pulmonary angiography. We have reported previously that serial noninvasive leg testing provides a practical noninvasive alternative to pulmonary angiography in patients with non-high probability (nondiagnostic) lung scans and adequate cardiorespiratory reserve. We have reevaluated this observation prospectively to (1) confirm or refute our original observation in an independent cohort and (2) to increase the numbers sufficiently to provide narrow confidence limits for the observed outcomes. A prospective comparative study in 1564 consecutive patients with suspected pulmonary embolism who underwent ventilation-perfusion lung scanning and objective testing for proximal-vein thrombosis. On long-term follow-up of 627 patients with the following characteristics: (1) abnormal, nondiagnostic lung scans, (2) not taking anticoagulant therapy, and (3) serial noninvasive test results negative for proximal-vein thrombosis, only 12 patients (1.9%; 95% confidence limits, 0.8% to 3.0%) had venous thromboembolism. By comparison, venous thromboembolism on follow-up occurred in four (0.7%) of 586 patients (95% confidence limits, 0.02% to 1.3%) with normal lung scans in whom anticoagulant therapy was withheld and in eight (5.5%) of 145 patients (95% confidence limits, 1.8% to 9.2%) with high probability lung scans who received anticoagulant therapy. Our findings indicate that the clinician has a practical noninvasive strategy in patients with adequate cardiorespiratory reserve and nondiagnostic lung scans that (1) avoids pulmonary angiography, (2) identifies patients with proximal-vein thrombosis who require treatment, and (3) avoids the need for treatment and further investigation in the majority of patients.

Evaluation of heartworm immunodiagnostic tests

Summary In this report, the use of appropriate statistical methods for the evaluation of heartworm immunodiagnostic tests is discussed. The evaluation of these tests is complicated by factors causing variation in sensitivity, specificity, accuracy, and predictive values of positive and negative test results. The primary sources of inconsistency are variation in the prevalence of heartworm infection among populations of dogs and the sensitivity of immunodiagnostic tests to various categories of heartworm infections (ie, patent, immune-mediated occult, unisex occult, and immature occult). Sample size (ie, number of dogs tested) affects the confidence limit values of sensitivity and specificity. At least 100 dogs should be used in each testing group (infected and uninfected) to generate values of sensitivity or specificity within reasonably narrow confidence limits. Use of more than 200 dogs in each testing group contributes little to further narrowing of confidence limits. The selection of appropriate statistical tests for comparison of tests or comparison of the sensitivity or specificity of a single diagnostic test to various categories of heartworm infections is critical. The McNemar paired χ2 test is appropriate for comparison of diagnostic tests, but it must be done by use of duplicate sera from each animal. A χ2 test of independence, or, in the case of a small sample size, the Fisher exact test, is appropriate for comparing the sensitivity or specificity of a single diagnostic test to various categories of heartworm infection.

Experience of the Belgian Society of Medical Oncology with single-administration 5 g/m2 ifosfamide with mesna as second- or third-line therapy in advanced breast cancer

In an ongoing phase II trial conducted in advanced breast cancer, we tested a therapy schedule consisting of continuous, 24-h infusion of 5 g/m2 ifosfamide (IFO) in 3 1 dextrose saline with mesna (MSN), repeated every 3 weeks until disease progression. Since September 1988, 16 heavily pretreated patients with advanced disease (11 with visceral lesions) considered refractory to standard chemotherapy (regimens always including cyclophosphamide) have been included. Objective partial remissions were observed in two cases (one in liver and one in soft-tissue and pleural lesions), and disease stabilization for at least 3 months occurred in four cases. No treatment-related death was recorded and tolerance was judged to be excellent (six cases) or acceptable in all instances. The haematological toxicity consisted mainly of transient leucopenia (nadirs evaluated by WHO scale as grade 3 in 43% and grade 4 in 29%), sometimes associated with thrombocytopenia (grade 3 in 7% and grade 4 in 7%). Other side effects included nausea and/or vomiting (grade 3-4 in 33%); worsening of preexisting alopecia (five cases); haemorrhagic cystitis (one case); mild, transient somnolence (two cases); and moderate fluid retention (two cases). We concluded that infusion of 5 g/m2 IFO over 24 h with MSN rescue might represent an acceptable second- or third-line salvage regimen. Close monitoring of haematological and renal function parameters is recommended. A larger number of patients will be treated in a continuation of this study to evaluate the true response rate within narrower confidence limits.

Sampling Statistics and Detection or Estimation of Diverse Populations of Stored-Product Insects1

Sampling data for a variety of stages and species of insects, sampling devices, locations in the marketing systems, and types of grain fit a single model with narrow confidence limits, indicating that the same sampling statistics might be used with diverse populations of stored-product insects. The progressively faster increase in variance and the progressively slower increase in the fraction of samples with insects as insect population density increased can both be explained by a purely probabilistic increase in the chance of a sample having more than one insect.

Prenatal measurement of the fetal cerebellum and cisterna cerebellomedullaris by ultrasound.

The fetal cerebellum can be visualized with ultrasound throughout the second trimester. We describe a technique for measuring the transverse and anteroposterior cerebellar diameters and the measurement of the cisterna magna in the same plane between 14 and 32 weeks gestation. Nomograms for these measurements against gestational age showed good correlation, and narrow confidence limits for the transverse cerebellar diameter. The transverse cerebellar diameter was also measured directly in 79 fetuses after midtrimester abortion and the measurements obtained were compared with the ultrasound TCD nomogram. Good correlation, was obtained between the post mortem measurements and the ultrasound TCD nomogram. Routine use of these measurements and nomograms should prove valuable in the diagnosis of congenital abnormality of the posterior fossa and may also be of use in assessing the effect of severe intrauterine growth retardation and other insults on cerebellar growth and development. The narrow confidence limits obtained with the TCD nomogram should enable it to be used with confidence in clinical practice.

Analysis of the activity of DNA, RNA, and protein synthesis inhibitors on Xenopus embryo development.

The teratogenic and growth-inhibiting potential of DNA, RNA, and protein synthesis inhibitors was explored using the Frog Embryo Teratogenesis Assay: Xenopus (FETAX). Endpoints measured in 96-h static tests were survival, malformation, ability to swim, skin pigmentation, stage of development, and growth. The DNA synthesis inhibitors hydroxyurea, cytosine arabinoside, and ethidium bromide proved to be teratogenic by the severity of malformations induced. Hydroxyurea gave an LC50 of 1.82 mg/ml, an EC50 (malformation) of 0.43 mg/ml, while the values for cytosine arabinsode were 5.41 and 0.76, respectively. The values for ethidium bromide were 0.05 and 0.035. The RNA synthesis inhibitor actinomycin D and the protein synthesis inhibitor cycloheximide were more embryolethal than teratogenic but significantly inhibited growth as determined by head-tail length measurements. Actinomycin D caused severe malformations, while cycloheximide caused relatively minor abnormalities. The LC50 for actinomycin D was 1.89 mg/ml, while the EC50 (malformation) was 2.17 mg/ml. For cycloheximide, the values were 1.59 and 1.19, respectively. FETAX advantages include rapid data collection, the ability to measure stage-dependent effects, and the ability to use a large number of embryos to obtain excellent dose-response curves with narrow confidence limits. Disadvantages include lack of a metabolic activation system, absence of a placental relationship, and the inability to detect specific abnormalities such as limb defects in 96 h.

Temperature-programmed reduction. Parametric sensitivity and estimation of kinetic parameters

The influence of the different experimental operating variables used in temperature-programmed reduction (TPR) measurements on the reduction profiles has been studied using the reduction of nickel oxide as an example. The temperature range and the shape of the resulting TPR profiles were markedly affected by the experimental conditions. A characteristic number, K, is defined to aid in selecting the values for the operating variables which should be chosen in order to obtain optimum reduction profiles. The number relates the heating rate, hydrogen concentration, total flow rate, and the amount of reducible sample in such a way that the operating variables can easily be adjusted. Upper and lower limits are proposed for this characteristic number, and the influence of its variation on the sensitivity of the TPR method is discussed. Different methods for estimating the kinetic parameters of the reduction of nickel oxide from TPR experiments are compared. A peak shape analysis leads to narrower confidence limits for the kinetic parameters than an estimation based on the shift of the temperature of the maximum reduction rate measured for different heating rates. Simple criteria are proposed to check whether TPR measurements are free from mass transfer limitations.

Reply to Levinton

Levinton notes that the number of individuals available to my study of patterns in the evolution of Homo erectus is low, especially when cranial base breadth is considered. If measurements of biauricular breadth are related to stratigraphic age of the fossils using linear regression, the slope obtained is small (−1.52). The confidence interval includes zero, and there is no evidence that a real trend is present. At the same time, the .95 interval is large (±26.81), so slopes differing by even a substantial margin cannot be distinguished. Levinton suggests that this negates a claim for stasis, and I agree that there are grounds for doubt. But by his reasoning, a case for stasis can never be established. While it is true that the occurrence of (very small) trends can never be denied, even if an analysis of many fossils shows a zero slope bracketed by narrow confidence limits, surely these “trends” must be dismissed as insignificant. Given such an outcome, a conclusion of no (directional) change would be quite justified.

Finding the Better of Two Similar Designs by Monte Carlo Techniques

During the iterative design of a system (or circuit) the designer is often faced with the problem of ranking two designs according to some criterion. If the system elements have uncertain values, each system manifestation can be evaluated as to whether or not it meets some performance criterion. The fraction which meets or exceeds the criterion is called the yield. Monte Carlo techniques can be used to simulate the population of systems and thus to estimate the ranking of two designs. The first result presented in the paper is a derivation of the probability that one design is better than another, along with confidence limits for that probability. If the two designs are topologically the same, i.e., they differ only in the nominal values and actual distributions of true parameter values, then the same set of random numbers can be used for one simulation of each design. Due to the similarity there may be a positive correlation between the 2 results which can then be used to narrow the confidence limits from the crude method mentioned above. The second result is a derivation of these narrow confidence limits.

Brunhes epoch geomagnetic secular variation on Marion Island: contribution to evidence for a long-term regional geomagnetic secular variation maximum.

A paleomagnetic study of 30 lavas from Marion Island, situated on the Indian-Atlantic ridge yields a value for Brunhes epoch secular variation with confidence limits so broad that no distinction can be made with present geomagnetic field behavior. When combined with similar results from the nearby Crozet Islands, however, data for over 100 separate lavas become available for analysis, facilitating an estimate of regional secular variation with narrow confidence limits. The result, which suggests that Brunhes epoch geomagnetic secular variation in this region was (at the 95 percent confidence level) higher than at present, could be explained if the present secular variation maximum south of South Africa persisted during at least the past half million years. It is concluded that values of Brunhes epoch geomagnetic secular variation may depend not only on latitude, experimental methods, and coverage of the geological interval involved, but also on proximity to long term isoporic foci. Incidental results for the survey include detection of excursions attributable to non-dipole behavior in some instances, and possible dipole tilting in other cases.

Effect of beta-receptor blocking drugs on the heart rate during submaximal exercise

The effects on the heart rate during exercise on a bicycle ergometer of increasing doses of various beta-receptor blocking drugs (dl-propranolol, dl-DU 21445, dl-DU 21523 and l-DU 21445) were examined. The drugs were given according to a triple cross-over design scheme. A linear relation existed between the decrease in heart rate during exercise and the logarithm of the dose, within the range studied. dl-DU 21445, a new beta-receptor blocking drug was 1.8 times more active after oral, and 3.2 times more active after intravenous administration, than dl-propranolol. The experimental design enabled narrow confidence limits to be obtained with a minimum of subjects. Exercise tests are a useful method of comparing beta-receptor blocking drugs in human subjects.