Performance Assessment of Coupled Tests: The Effects of Statistical Non-Independence

Chiecchio, Andrea; Malvano, R.; Giglioli, F.R.; Bo, Adedokun

doi:10.1515/cclm.1994.32.3.169

Cited by 4 publications

(3 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In parallel testing, if either 1 of 2 tests meets the respective test-positive criterion, the subject is considered test positive (eg, if cutoff A for wet or cutoff B for soiled is met, the subject is test positive). 9 In addition, the cutoff that captured approximately 90% of the true cases of excess weight loss was identified. Confidence intervals (95%) were constructed for each Se and Sp estimate, and Youden's J statistic was calculated for each Se/Sp pair.…”

Section: Discussionmentioning

confidence: 99%

Newborn Wet and Soiled Diaper Counts and Timing of Onset of Lactation as Indicators of Breastfeeding Inadequacy

et al. 2008

View full text Add to dashboard Cite

Using data from 242 exclusively breastfeeding mother-infant pairs, newborn elimination patterns were analyzed. Sensitivity (Se) and Specificity (Sp) of day 4 (72-96 hours) wet and soiled output, in addition to the timing of onset of lactation, in identifying cases of breastfeeding inadequacy (defined as neonatal weight loss >or= 10% of birth weight) were examined. The usefulness of 2 measures in parallel was also explored. Median number of diapers (wet, soiled) on days 1, 4, and 7, respectively, was 2, 3; 5, 4; and 7, 6. The most efficient day 4 predictor of breastfeeding inadequacy was soiled diaper output or= 72 hours); Se = .86 (95% confidence interval, .73-.99) and Sp = .59 (.55-.63). Fewer than 4 soiled diapers on day 4 when used in conjunction with delayed onset of lactation may be indicative of breastfeeding inadequacy, but low specificity will result in many false positives.

show abstract

Section: Discussionmentioning

confidence: 99%

Newborn Wet and Soiled Diaper Counts and Timing of Onset of Lactation as Indicators of Breastfeeding Inadequacy

et al. 2008

View full text Add to dashboard Cite

show abstract

“…Then, combinations of components can be systematically designed, making, for example, use of decision theoretical tools, whenever possible also including analysis of conditional dependencies of tests. 141,147,148 The simplest example, ie, to test or not to test, has been addressed above. Information required to make this decision is the performance of the test, ie, sensitivity and specificity, the prevalence in the test population, the cost of misclassifications, ie, for false negative and false positive, and the cost of the test.…”

Section: Composition Of An Integrated Testing Strategy (Its)mentioning

confidence: 99%

Toward an evidence-based toxicology

2006

View full text Add to dashboard Cite

The increasing demands on toxicology of large-scale risk assessment programmes for chemicals and emerging or expanding areas of chemical use suggest it is timely to review the toxicological toolbox. Like in clinical medicine, where an evidence-based medicine (EBM) is critically reviewing traditional approaches, toxicology has the opportunity to reshape and enlarge its methodology and approaches on the basis of compounded scientific knowledge. Such revision would have to be based on structured reviews of current practice, ie, assessment of test performance characteristics, mechanistic understanding, extended quality assurance, formal validation and the use of integrated testing strategies. This form of revision could optimize the balance between safety, costs and animal welfare, explicitly stating and, where possible, quantifying uncertainties. After a self-critical reassessment of current practices and evaluation of the thus generated information, such an evidence-based toxicology (EBT) promises to make better use of resources and to increase the quality of results, facilitating their interpretation. It shall open up hazard and also risk assessments to new technologies, flexibly accommodating current and future mechanistic understanding. An EBT will be better prepared to answer the continuously growing safety demands of modern societies.

show abstract

“…In their paper Chiecchio et al (1) corisider combinations of the results of two laboratory tests for diagnostic decisipns according to the AND and OR rule. With the AND rule a case is classified äs pathologic if both tests classify this case äs pathologic.…”

Section: Introductionmentioning

confidence: 99%

Combining Laboratory Tests for Diagnostic Decisions

Schneider¹

1994

CCLM

View full text Add to dashboard Cite

Summary:For the combination of two diagnostic tests often the AND and OR procedure is used. Üie AND proeedure classifies a case äs pathologic, if both tests classify it äs pathologic; the OR procedure, if at least one test classifies it äs pathologic. It is shown that these are no good combination procedures resulting in diagnostic qualities which can be even worse than the worst single test. Instead optimal multivariate Statistical classification procedures should be used which give the best possible improvement of diagnostic quality.

show abstract

Performance Assessment of Coupled Tests: The Effects of Statistical Non-Independence

Cited by 4 publications

References 9 publications

Newborn Wet and Soiled Diaper Counts and Timing of Onset of Lactation as Indicators of Breastfeeding Inadequacy

Newborn Wet and Soiled Diaper Counts and Timing of Onset of Lactation as Indicators of Breastfeeding Inadequacy

Toward an evidence-based toxicology

Combining Laboratory Tests for Diagnostic Decisions

Contact Info

Product

Resources

About