Detection bias

Three methods were applied to detect bias measurements in this process network where only total mass balances are considered. These are the Generalized Ratio... 

For determination of minimum detectable bias (MDB) at the 95% confidence level ... 

Many diseases are "silent," and people may not realize they are ill. Detection bias can exist if all subjects do not receive the same medical scrutiny. In retrospective case-control studies, detection bias occurs if the "cases" receive more intensive medical scrutiny than the "controls" and silent diseases in the control group more often remain undetected. 

Detection bias occurs in convenience-cohort studies when any measure of substance exposure is correlated with differences in medical scrutiny. For example, the positive relationship between reser-pine (a blood pressure medication) and breast cancer might be attributable to the fact that women under treatment for high blood pressure are more likely to have breast exams, which detect otherwise silent breast cancers (Feinstein 1988, 1261). The same might be true of the relationship between alcohol intake and breast cancer, because alcohol could be a surrogate for income and more frequent breast cancer screening and mammography (Feinstein 1988, 1261). 

Case-control and convenience-cohort studies are susceptible to unobserved differences between experimental- and control-group members. This leads to selection and detection bias, which accounts for the alternative explanations of the observed statistical relationship between exposure and disease incidence. The best individual-level epidemiological studies go to great lengths to avoid bias, use large numbers of people, and study large exposures. 

Feinstein s arguments about detection bias come from his comparison of data on disease incidence reported by hospitals and pubhc health agencies and incidence data derived from autopsies. While his arguments have generated much controversy among epidemiologists, I believe his responses are more than adequate. See Kass and Shapiro (1989), Feinstein (1989), Poole et al. (1990), and Feinstein et al. (1990). 

Much of the work on both sides of the argument was less than watertight. Some studies failed to consider the confounding effects of other risk factors (notably smoking) or the likelihood of detection bias (particularly for venous thromboembolism, which is much more common in young women than myocardial infarction or stroke). The results of some studies were also confounded by uncertainties in the history of drug exposure. A landmark paper to... 

Precision Instrument repeatability - ten replicate injections. RSD<1% Intra-assay precision. Multiple analysis of aliquots of a sample during one day. RSD < 2% Intermediate precision. Multiple analysts, instruments, days in one laboratory. Reproducibility by inter-laboratory studies to detect bias. 

External quality assurance (EQA) is fundamental to the standardization of clinical laboratory methods because it provides a means to compare results generated in one laboratory with those of peer laboratories subscribing to the same EQA program. EQA programs are especially beneficial since internal QA and QC procedures are limited in their ability to detect bias in analytical methods. Internal QA/QC can only detect errors that result in a deviation from the original method validation inherent errors in the method may go unnoticed. Therefore, it is helpful to compare the results produced by a new method with those from other laboratories (Burtis and Ashwood, 2001). Monitoring the performance of laboratory procedures in a consistent manner keeps the laboratory accountable, and can reveal systematic errors that would otherwise be undetected. A prominent component of EQA is proficiency testing. 

Mixtures of several elements or substances in solvents are available from many suppliers in certified quality. Such materials are very useful to environmental monitoring as many represent excellent materials for the calibration of instruments. Several target contaminants like PCB, PAH, PCDD and PCDF, metals etc. are available from NIST, BCR etc. Such certified materials have purity figures with small uncertainties, they can be used without any particular precaution compared to the equivalent pure substance. They are valuable tools to detect bias in calibration and allow rapid correction. Uriano and Gravatt [8] have cited an example of the use of CRMs to correct for bias in SO measurements in air and in particular how CRMs can help to correct for additive or multiplicative bias in interlaboratory studies. Massart et al. have also discussed similar effects on signals of detectors [9] and the reader should refer to them. 

The external validity refers to applicability and generalization and is outlined in the section, Applying the Results. The remainder of this section focuses on critically appraising the quality— that is, the internal validity—of individual trials. The internal validity is determined by how well the trial ensures that the known and unknown risk factors are equally distributed between the treatment and control groups. To ensure validity, the conduct of the trial should minimize systematic bias and random error as much as possible to provide results that are as accurate and close to the truth as possible. Four sources of bias are possible in trials of health care interventions selection bias, performance bias, attrition bias, and detection bias. Bias can result in an overestimation or underestimation of the effectiveness of a drug therapy and mislead the reader. While it is beyond... 

Hypothesis testing is applicable to all of the above factors. Detection decisions may be made, for example, using the critical level of Student s-t to test for bias, or the critical level of to test an assumed spectral shape or calibration model or error model. For a given measurement design and assumption test procedure, one can estimate the corresponding detection limit for the alternative hypothesis, e.g., the minimum detectable bias. As with analyte detection, the ability to detect erroneous assumptions rests heavily on the design of the experiment and the study of optimal designs is a field unto Itself. 

Several calculations were done to compare experimental data with the results of KENO calculations. The experimental data were of two types (a) assemblies of plutonium metal disks interleaved with and reflected by Plexiglas, and (b) water-flooded and -reflected arrays of metal cubes or rods. The agreement between computed and measured critical masses was satisfactory, ranging from a 1.6 to a 6.2% discrepancy with no detectable bias. 

Coronary artery disease Treatment with nitrates and calcium antagonists, which decrease the LES pressure detection bias with chest pain evaluation... 

Duodenal ulcer Detection bias of upper gastrointestinal endoscopy... 

Samples containing heavy metal additives, lead alkyls, etc., may interfere with the test method. Elements such as silicon, phosphorus, calcium, potassium, and halides interfere if present in concentrations of more than a few hundred milligrams per kilogram. Consult instrument manufacturers instructions for specific interference data. Materials used in the study to determine precision contained up to 250 mg/kg vanadium, 50 mg/kg nickel, and 15 mg/kg iron with no detectable bias introduced. 

---