Evaluating Systematic Biases

In addition to chance, systematic biases can also affect the relationship between an exposure and disease. Biases lead to an incorrect estimate of the relationship between the exposure and disease that is an incorrect measure of the relative risk. Some biases will result in an effect being observed (i.e., statistically significant RR) when there is not a causal relationship, whereas other biases will result in obscuring a causal relationship between exposure and disease (refer to as biasing toward the null hypothesis). In an individual study, biases can be introduced during the selection of the subjects, follow-up of disease status, or exposure assessment. Biases can also occur in the evaluation of a causal relationship across studies. 

---

Figure 5.25 Predicted vs. reference concentration plot and several related statistics to evaluate the adequacy of the PLS model (original data from Figure 5.9, CS2 example, autoscaled, 4 factors in the PLS model).The theoretical line is displaced to simplify its visualization and no systematic bias was present in the predictions.

As discussed in Chapter 6, this principle is termed "The Fimdamental Attribution Error." It contributes to systematic bias whenever we attempt to evaluate others, from completing performance appraisals to conducting an injury investigation. Because we are quick to attribute internal (person-based) factors to other people s behavior, we tend to presume consistency in others because of permanent traits or personality characteristics. To explain injuries to other persons, we use expressions like, "He s just careless," "She had the wrong attitude," and "They were not thinking like a team."... 

We will begin by taking a look at the detailed aspects of a basic problem that confronts most analytical laboratories. This is the problem of comparing two quantitative methods performed by different operators or at different locations. This is an area that is not restricted to spectroscopic analysis many of the concepts we describe here can be applied to evaluating the results from any form of chemical analysis. In our case we will examine a comparison of two standard methods to determine precision, accuracy, and systematic errors (bias) for each of the methods and laboratories involved in an analytical test. As it happens, in the case we use for our example, one of the analytical methods is spectroscopic and the other is an HPLC method. 

A systematic evaluation of the biological position of living things started with Linnaeus in the 18 centxuy, albeit limited, with few exceptions, to conspicuous species. This lopsided bias has continued down to our days, judging from the meager number of taxonomists devoted to microorganisms, estimated to a mere 2-3% of the whole community of taxonomists (May 1994). 

As already mentioned in the introduction, ruggedness is a part of the precision evaluation. Precision is a measure for random errors. Random errors cause imprecise measurements. Another kind of errors that can occur are systematic errors. They cause inaccurate results and are measured in terms of bias. The total error is defined as the sum of the systematic and random errors. 

The error of an analytical result is related to the (in)accuracy of an analytical method and consists of a systematic component and a random component [14]. Precision and bias studies form the basis for evaluation of the accuracy of an analytical method [18]. The accuracy of results only relates to the fitness for purpose of an analytical system assessed by method validation. Reliability of results however has to do with more than method validation alone. MU is more than just a singlefigure expression of accuracy. It covers all sources of errors which are relevant for all analyte concentration levels. MU is a key indicator of both fitness for purpose and reliability of results, binding together the ideas of fitness for purpose and quality control (QC) and thus covering the whole QA system [4,37]. 

The measurement of the recoveries of analyte added to matrices of interest is used to measure the bias of a method (systematic error) although care must be taken when evaluating the results of recovery experiments as it is possible to obtain 100% recovery of the added standard without fully extracting the analyte which may be bound in the sample matrix. 

The TACAN Corporation has evaluated the stability of drive voltage and bias voltage under conditions of extended operation and for irradiation with light of different wavelengths and optical power levels [306, 307]. The performance of polymer modulators was reasonably good and systematically improved as lattice hardness increased (see Figs. 6,20 and 34). We have observed comparable stability. 

The accuracy of exposure assessment is determined by systematic and random errors in the assessment. For quantitative exposure assessments, important sources of error include measurement errors (i.e. from laboratory and field monitoring techniques), as well as variations in exposure over time and space. For qualitative exposure proxies (e.g. self-reported past exposures, occupational histories or expert evaluations), the most important sources of error are recall bias (systematic differences in exposure recall between cases and controls) and random error, expressed in terms of intra- and inter-rater agreement. Although systematic errors can result in serious misinterpretations of the data, especially due to scaling problems, random errors have received more attention in epidemiology because this type of error is pervasive, and its effect is usually to diminish estimates of association between exposure and disease. The magnitude of random errors can be considerable in epidemiological field studies. 

The evaluation study will determine the attributes (bias, precision, specificity, limits of detection) of the immunoassay. Bias testing (systematic error) will be conducted by measuring recoveries of the analyte added to matrices of interest. Replicate analysis will be performed on blind replicates or split levels (e.g., Youden pairs). A minimum number of replicates will be performed to provide statistically meaningful results. The number of replicates will be determined by the intended purpose of the immunoassay as well as the documented method performance of the comparative method. 

Accuracy is the degree of agreement of a measured value with the true value of the quantity under concern. Inaccuracy results from imprecision (random error) and bias (systematic error) in the measurement process. Bias can only be estimated from the results of measurements of samples of known composition. SRMs are ideal for use in such an evaluation (22). 

We undertook this review and evaluation with the intent of providing the reader a resource to access original literature published assessing the economic value of clinical pharmacy services, and to evaluate the quality of that literature. The articles included in this review represent only those published in standard literature. We did not consider unpublished studies and therefore our results may be subject to inherent publication bias (so-called file drawer effect). We included only articles that contained some consideration of the financial impact of clinical pharmacy services. Certainly, many useful articles describe and evaluate clinical pharmacy services, but focus on nonfinancial outcomes and impact, and are worthy of review. Finally, our review of the literature, although intended to be systematic and thorough, may not have captured all the published literature on this topic. 

The objective of any review of experimental values is to evaluate the accuracy and precision of the results. The description of a procedure for the selection of the evaluated values (EvV) of electron affinities is one of the objectives of this book. The most recent precise values are taken as the EvV. However, this is not always valid. It is better to obtain estimates of the bias and random errors in the values and to compare their accuracy and precision. The reported values of a property are collected and examined in terms of the random errors. If the values agree within the error, the weighted average value is the most appropriate value. If the values do not agree within the random errors, then systematic errors must be investigated. In order to evaluate bias errors, at least two different procedures for measuring the same quantity must be available. 

Further aspects, pros and cons of WPPF, are discussed in Chapter 5. Here it is important to underline the fact that the validity of profile fitting is limited by the basic assumption of using an a priori selected profile function without any sound hypothesis that the specific functional form is appropriate to the case of study. The consequence of this arbitrary assumption can be quite different. For example, in most practical cases, profile fitting can provide reliable values of peak position and area, whereas the effects on the profile parameters are less known and rarely considered. The arbitrary choice of a profile function tends to introduce systematic errors in the width and shape parameters, which invariably introduce a bias in a following LPA, whose consequences can hardly be evaluated. It is therefore a natural tendency, for complex problems and to obtain more reliable results, to remove the a priori selected profile functions - leading to the following section dedicated to Whole Powder Pattern Modelling methods. 

Was the study double-blinded To minimize performance bias (systematic differences in the care provided, apart from the intervention being evaluated), the subjects and the clinicians should be unaware of the therapy received. The double-blind... 

Method precision (random error, variation) and accuracy (systematic error, mean bias) for LBAs should be evaluated by analyzing validation samples (QC samples) that are prepared in a biological matrix that is judged scientifically to be representative of the anticipated study samples [18]. This topic has been reviewed in other publications [3 6,9,10,20]. These performance characteristics should be evaluated during the method development phase, taking into consideration the factors known to vary in the method (e.g., analysts, instruments, reagents, different days, etc.). Several concentrations are required during the method development phase and are assayed in replicates. Factors known to vary between runs (e.g., analyst, instrument, and day)... 

Given a set of basis states, excited eigenstates can be computed varia-tionally by solving a linear variational problem, and the Metropolis method can be used to evaluate the required matrix elements. The methods involving the power method, as described above, can then be used to remove the variational bias systematically [13,19,20]. 

It can be assumed that with the development and study of new methods, the ability to determine M (S), the method bias component of uncertainty, cannot be done given that it can be evaluated only relative to a true measure of analyte concentration. This can be achieved by analysis of a certified reference material, which is usually uncommon, or by comparison to a well-characterized/accepted method, which is unlikely to exist for veterinary drug residues of recent interest. Given that method bias is typically corrected using matrix-matched calibration standards, internal standard or recovery spikes, it is considered that the use of these approaches provides correction for the systematic component of method bias. The random error would be considered part of the interlaboratory derived components of uncertainty. 

---