Contamination Bias Research Articles

Prediction of spurious HLA class II typing results using probabilistic classification

While modern high-throughput sequence-based HLA genotyping methods generally provide highly accurate typing results, artefacts may nonetheless arise for numerous reasons, such as sample contamination, sequencing errors, read misalignments, or PCR amplification biases. To help detecting spurious typing results, we tested the performance of two probabilistic classifiers (binary logistic regression and random forest models) based on population-specific genotype frequencies. We trained the model using high-resolution typing results for HLA-DRB1, DQB1, and DPB1 from large samples of German, Polish and UK-based donors. The high predictive capacity of the best models replicated both in 10-fold cross-validation for each gene and in using independent evaluation data (AUC 0.820–0.893). While genotype frequencies alone provide enough predictive power to render the model generally useful for highlighting potentially spurious typing results, the inclusion of workflow-specific predictors substantially increases prediction specificity. Low initial DNA concentrations in combination with low-volume PCR reactions form a major source of stochastic error specific to the Fluidigm chip-based workflow at DKMS Life Science Lab. The addition of DNA concentrations as a predictor variable thus substantially increased AUC (0.947–0.959) over purely frequency-based models.

Management of aromatase inhibitor-associated musculoskeletal symptoms: A systematic review.

157 Background: Aromatase inhibitor-associated musculoskeletal symptoms (AIMSS) are experienced by a significant number of women on adjuvant aromatase inhibitors for breast cancer. These symptoms can result in therapy noncompliance and there is no accepted standardised treatment. We believe that this is the first systematic review to consider the evidence for all pharmacological and non-pharmacological interventions in the treatment of AIMSS. Methods: We conducted a systematic search in PubMed, EMBASE, CINAHL and CENTRAL. Clinical trials and observational studies for all potential pharmacological and non-pharmacological interventions were included. Two reviewers screened the results and extracted the relevant data. Risk of bias for the full papers were assessed using JADAD or a modified Newcastle Ottawa score, with conference abstracts and letters to the editor excluded from this process. Results: Of 1260 records being identified, 887 unique citations were screened with 93 full text papers retrieved for assessment, and 37 selected for inclusion. We identified 9 pharmacological interventions (RCT = 3, pre-post = 4, cohort = 2) and 28 non-pharmacological interventions (RCT/CCT = 16, pre-post = 8, cohort = 3). The largest number of studies were for various exercise interventions (n = 7) with the largest trial showing a 29% decrease in worst joint pain scores compared to the control (2%, p < 0.001). The highest number of RCTs was for acupuncture (n = 5), the majority of which showed a significant benefit with both real and sham acupuncture. Other single RCTs showing benefit include the use of testosterone, calcitonin and etoricoxib. No study adequately controlled for contamination bias from extraneous variables of both pharmacological and non-pharmacological interventions. Conclusions: Given that pharmacological treatment is often recommended for AIMSS, it is surprising that there is limited published evidence for its use. Although the interventions being used appear tolerable with minimal adverse effects, the current level of evidence is low, and additional large RCTs with more rigorous controlling for contamination from other interventions are required to confirm some of the reported promising results.

Balancing Contamination and Referral Bias in a Randomized Clinical Trial: An Application of Pseudo-Cluster Randomization.

In randomized trials of provider-focused clinical interventions, treatment allocation often cannot be blinded to participants, study staff, or providers. The choice of unit of randomization (patient, provider, or clinic) entails tradeoffs in cost, power, and bias. Provider- or clinic-level randomization can minimize contamination, but it incurs the equally problematic potential for referral bias; that is, because arm assignment of future participants generally cannot be concealed, differences between arms may arise in the types of patients enrolled. Pseudo-cluster randomization is a novel study design that balances these competing validity threats. Providers are randomly assigned to an imbalanced proportion of intervention-arm participants (e.g., 80% or 20%). Providers can be masked to the imbalance, avoiding referral bias. Contamination is reduced because only a minority of control-arm participants are treated by majority-intervention providers. Pseudo-cluster randomization was implemented in a randomized trial of a decision support intervention to manage depression among patients receiving human immunodeficiency virus care in the southern United States in 2010-2014. The design appears successful in avoiding referral bias (participants were comparable between arms on important characteristics) and contamination (key depression treatment indicators were comparable between usual care participants managed by majority-intervention and majority-usual care providers and were markedly different compared with intervention participants).

Assessment of contamination and misclassification biases in a randomized controlled trial of a social network peer education intervention to reduce HIV risk behaviors among drug users and risk partners in Philadelphia, PA and Chiang Mai, Thailand.

Controlled trials of HIV prevention and care interventions are susceptible to contamination. In a randomized controlled trial of a social network peer education intervention among people who inject drugs and their risk partners in Philadelphia, PA and Chiang Mai, Thailand, we tested a contamination measure based on recall of intervention terms. We assessed the recall of test, negative and positive control terms among intervention and control arm participants and compared the relative odds of recall of test versus negative control terms between study arms. The contamination measures showed good discriminant ability among participants in Chiang Mai. In Philadelphia there was no evidence of contamination and little evidence of diffusion. In Chiang Mai there was strong evidence of diffusion and contamination. Network structure and peer education in Chiang Mai likely led to contamination. Recall of intervention materials can be a useful method to detect contamination in experimental interventions.

Balancing Contamination and Referral Bias in a Randomized Clinical Trial: An Application of Pseudo-Cluster Randomization

Balancing Contamination and Referral Bias in a Randomized Clinical Trial: An Application of Pseudo-Cluster Randomization

Effects of dietary weight loss on obstructive sleep apnea

1. The data extraction on apnea–hypopnea index (AHI) from the studies included in this meta-analysis seems to have a selection bias. The authors included the study by Schwartz et al. [2] which only reported the NREM AHI, and no REM AHI was reported for their 13 study participants. Similarly, the study by Smith et al. [3] reported separate AHI data for NREM and REM sleep. However, the authors of this meta-analysis selectively included the NREM AHI data and left out the REM AHI data. 2. There also appears to be a contamination bias in the baseline characteristics of the study populations of included studies. Two of their included studies [4, 5] included study participants that were concurrently on continuous positive airway pressure (CPAP). In the study by Foster et al. [4], at least 5 % of the included participants were on CPAP. Similarly, in the study by Nerfeldt et al. [5], of the 33 study participants who started the study, 19 were on CPAP. Although in both studies polysomnographic assessments of AHI were performed without the CPAP, the study population of these two studies does not match with the rest of the included studies in this meta-analysis in which the sole intervention was dietary weight loss. 3. In their analysis of the preto postintervention on body mass index and AHI, it is unclear why the authors chose to use the Mantel–Haenszel method of meta-analysis. This method is used for the meta-analysis of dichotomous outcomes, and generally, an inverse-variance method is used for the continuous variables. This is an incorrect way of analyzing paired data of means and standard deviations (SD). 4. There is a major statistical error in analyzing the AHI data from the study by Foster et al. [4] in Fig. 4. For the difference in AHI between control and intervention, the authors used a mean AHI of −5.4 and SD of 1.5 for the intervention group and a mean AHI of 4.2 and SD of 1.4 for the control group. Whereas, Foster and colleagues clearly stated that 1.5 and 1.4 were actually the standard errors and not SD. It is incorrect to combine SDs with standard errors in a meta-analysis. 5. Similarly, in Figure 4, it is not clear as to how the authors calculated the SD of difference for the difference in AHI between control and intervention groups for the study by Smith et al [3].

Timing in multitasking: Memory contamination and time pressure bias

There can be systematic biases in time estimation when it is performed in complex multitasking situations. In this paper we focus on the mechanisms that cause participants to tend to respond too quickly and underestimate a target interval (250–400ms) in a complex, real-time task. We hypothesized that two factors are responsible for the too-early bias: (1) Memory contamination from an even shorter time interval in the task, and (2) time pressure to take appropriate actions in time. In a simpler experiment that was focused on just these two factors, we found a strong too-early bias when participants estimated the target interval in alternation with a shorter interval and when they had little time to perform the task. The too-early bias was absent when they estimated the target interval in isolation without contamination and time pressure. A strong too-late bias occurred when the target interval alternated with a longer interval and there was no time pressure to respond. The effects were captured by incorporating the timing model of Taatgen and van Rijn (2011) into the ACT-R model for the Space Fortress task (Bothell, 2010). The results show that to properly understand time estimation in a dynamic task one needs to model the multiple influences that are occurring from the surrounding context.

South American immigrants in the USA

PurposeThe purpose of this paper is to examine the main factors influencing the decision of migrants from South America (SA) living in the USA to return to their countries of origin. In particular, it is aimed at understanding the role played by higher education in this decision. Although it is not feasible to refer to the tasks performed by immigrants in the USA as determinants in their decision to return to their homelands, it is nonetheless worth verifying whether these tasks are closely related or not to their qualifications.Design/methodology/approachThe program evaluation methodology proposed by Heckman and Robb to correct the “contamination bias” in a program evaluation setting is adopted in this document.FindingsThis exercise shows that the probability of remaining in the USA for those with university degrees or more advanced education levels is 12.6 percent higher than for those with only secondary or lower education. In addition, it is shown that SA immigrants are usually engaged in tasks requiring certain qualifications commensurate with their level of education, although such relationship varies in intensity among migrants from different countries.Research limitations/implicationsThis exercise does not take into account migrants in the USA not included in the US Censuses.Practical implicationsThe results obtained mean that the flight of human capital from SA countries to the USA is being accentuated by the “negative selection” of returnees, and that migrants' qualifications actually determine the complexity of tasks they may perform in the USA.Originality/valueThe paper overcomes the lack of longitudinal data of migrants by implementing a methodology providing new evidence about return migration and brain drain of SA immigrants in the USA. In addition, new data are used to analyze the relationship between education and the complexity of the tasks SA migrants are performing in the USA.

Verbal fluency deficits in clinically isolated syndrome suggestive of multiple sclerosis

BackgroundLittle is known about relationships between Cognitive Impairment (CI) and verbal fluency measures in patients with a Clinically Isolated Syndrome (CIS) suggestive of Multiple Sclerosis. The aim of this study was to assess the extent of verbal fluency deficits and their predictive value for the presence of a CI in a population of CIS patients. MethodsCI was detected by the Rao's Brief Repeatable Battery (BRB) and the Stroop Test (ST) in 100 CIS patients. The BRB includes the Word List Generation (WLG) test for semantic verbal fluency. The FAS test was used to investigate phonemic verbal fluency. CI was defined as the failure in at least 3 tests on BRB (without WLG to exclude criterion contamination bias) and ST. ResultsEleven patients failed in at least 3 of the BRB tests and were classified as cognitively impaired. The comparison of Receiver Operating Characteristic curves showed that the Area Under the Curve (AUC) of the WLG was not significantly different from the AUC of the FAS (0.787 vs 0.755; p=N.S.). A cut-off <17 words for the WLG achieved 64% of sensitivity and a 79% of specificity, and a cut-off <28 words achieved 82% of sensitivity and a 66% of specificity in discriminating patients with CI. ConclusionsVerbal fluency deficits occur early in the disease course and may predict the presence of CI in CIS patients.

A self-management program for employees with complaints of the arm, neck, or shoulder (CANS): study protocol for a randomized controlled trial.

BackgroundComplaints of the arm, neck, or shoulder (CANS) have a multifactorial origin and cause considerable work problems, including decreased work productivity, sickness absence, and, ultimately, job loss. There is a need for intervention programs for people with CANS. Self-management is an approach used in chronic disease care to improve self-efficacy and wellness behaviors to facilitate participants to make informed choices and carry them out. This study will evaluate the effectiveness of a self-management program (including ehealth) and compare it to usual care among employees with chronic CANS (lasting >3 months).Methods/designThis is a randomized controlled trial in which 142 participants will be recruited and randomized (with pre-stratification) to either the intervention group (IG) or control group (CG). The IG will participate in a self-management program consisting of six group sessions and an ehealth module. The CG is allowed to use all usual care available. The primary outcome of the study is the self-reported disability of arm, shoulder, and hand, measured with the Disabilities of the Arm, Shoulder and Hand questionnaire (DASH). Secondary outcomes include: absenteeism, pain in the previous week, quality of life, catastrophizing pain, self-efficacy, workstyle, presenteeism, fatigue, the use of usual care, and limitations experienced on the job. Data are collected at baseline and at 3, 6, and 12 months follow-up.DiscussionFollowing the process of intervention mapping we developed a self-management program to suit and alleviate the problems and needs of employees with CANS. A strength of the study is that our intervention is specifically tailored to match the needs of employees with CANS. The study also has some potential weaknesses (for example, use of co-interventions, combination of group sessions and ehealth, self-reporting of data and possible contamination, Hawthorne effect, and recall or information bias) which are discussed.Trial registrationThe trial is registered with the Dutch Trial Register (http://www.trialregister.nlNTR3816): (January 2013). The first participant was randomized in September 2012.

Excellent adherence and no contamination by physiotherapists involved in a randomized controlled trial on reactivation of COPD patients: a qualitative process evaluation study

ObjectiveTo assess the adherence of physiotherapists to the study protocol and the occurrence of contamination bias during the course of a randomized controlled trial with a recruitment period of 2 years and a 1-year follow-up (COPE-II study).Study design and settingIn the COPE-II study, intervention patients received a standardized physiotherapeutic reactivation intervention (COPE-active) and control patients received usual care. The latter could include regular physiotherapy treatment. Information about the adherence of physiotherapists with the study protocol was collected by performing a single interview with both intervention and control patients. Patients were only interviewed when they were currently receiving physiotherapy. Interviews were performed during two separate time periods, 10 months apart. Nine characteristics of the COPE-active intervention were scored. Scores were converted into percentages (0%, no aspects of COPE-active; 100%, full implementation of COPE-active).ResultsFifty-one patients were interviewed (first period: intervention n = 14 and control n = 10; second period: intervention n = 18 and control n = 9). Adherence with the COPE-active protocol was high (median scores: period 1, 96.8%; period 2, 92.1%), and large contrasts in scores between the intervention and control group were found (period 1: 96.8% versus 22.7%; period 2: 92.1% versus 25.0%). The scores of patients treated by seven physiotherapists who trained patients of both study groups were similar to the scores of patients treated by physiotherapists who only trained patients of one study group.ConclusionThe adherence of physiotherapists with the COPE-active protocol was high, remained unchanged over time, and no obvious contamination bias occurred.

Calling amplified haplotypes in next generation tumor sequence data

During tumor initiation and progression, cancer cells acquire a selective advantage, allowing them to outcompete their normal counterparts. Identification of the genetic changes that underlie these tumor acquired traits can provide deeper insights into the biology of tumorigenesis. Regions of copy number alterations and germline DNA variants are some of the elements subject to selection during tumor evolution. Integrated examination of inherited variation and somatic alterations holds the potential to reveal specific nucleotide alleles that a tumor "prefers" to have amplified. Next-generation sequencing of tumor and matched normal tissues provides a high-resolution platform to identify and analyze such somatic amplicons. Within an amplicon, examination of informative (e.g., heterozygous) sites deviating from a 1:1 ratio may suggest selection of that allele. A naive approach examines the reads for each heterozygous site in isolation; however, this ignores available valuable linkage information across sites. We, therefore, present a novel hidden Markov model-based method-Haplotype Amplification in Tumor Sequences (HATS)-that analyzes tumor and normal sequence data, along with training data for phasing purposes, to infer amplified alleles and haplotypes in regions of copy number gain. Our method is designed to handle rare variants and biases in read data. We assess the performance of HATS using simulated amplified regions generated from varying copy number and coverage levels, followed by amplicons in real data. We demonstrate that HATS infers the amplified alleles more accurately than does the naive approach, especially at low to intermediate coverage levels and in cases (including high coverage) possessing stromal contamination or allelic bias.

Contamination and harm relevant UCS‐expectancy bias in spider phobic individuals: influence of treatment

Phobic individuals expect aversive UCS's following encounters with phobic stimuli. Previous research using a thought-experiment procedure showed that contamination rather than harm-related outcome expectancies differentiated best between high and low spider fearful undergraduates. This study investigated the alleged role of these UCS-expectancy biases in the maintenance of phobic complaints. First, this study sought to replicate these earlier findings in a community sample of high spider fearful individuals who applied for treatment (n = 60) and a sample of low spider fear controls (n = 30). Second, the present study tested if UCS-expectancies disappear following successful treatment and whether there were any differences between harm and contamination-related UCS expectancies in this respect. If contamination- and/or harm-related UCS-expectancy biases play a critical role in the maintenance of spider fear, these biases should be substantially reduced after successful treatment. The results showed that spider fearful individuals associated spiders relatively strongly with both harm- and contamination-related outcomes. Consistent with the alleged reciprocal relationship between phobic fear and UCS expectancy bias, both types of biased expectancies were effectively reduced following treatment.

CHARACTERIZING THE CONTAMINATING DISTANCE DISTRIBUTION FOR BAYESIAN SUPERNOVA COSMOLOGY

Measurements of the equation of state of dark energy from surveys of thousands of Type Ia Supernovae will be limited by spectroscopic follow-up and must therefore rely on photometric identification, increasing the chance that the sample is contaminated by core collapse supernovae (CC SNe). Bayesian methods for SN cosmology can remove contamination bias while maintaining high statistical precision but are sensitive to the choice of parameterization of the contaminating distance distribution. We use simulations to investigate the form of the contaminating distribution and its dependence on the absolute magnitudes, light curve shapes, colors, extinction, and redshifts of CC SNe. We find that the CC luminosity function (LF) dominates the distance distribution function, but its shape is increasingly distorted as the redshift increases and more CC SNe fall below the survey magnitude limit. The shapes and colors of the CC light curves generally shift the distance distribution, and their effect on the CC distances is correlated. We compare the simulated distances to the first year results of the SDSS-II SN survey and find that the SDSS distance distributions can be reproduced with simulated CC SNe that are ∼1 mag fainter than the standard Richardson et al. LFs, which do not produce a good fit. To exploit the full power of the Bayesian parameter estimation method, parameterization of the contaminating distribution should be guided by the current knowledge of the CC LFs, coupled with the effects of the survey selection and magnitude limit, and allow for systematic shifts caused by the parameters of the distance fit.

Pseudo Cluster Randomization: Balancing the Disadvantages of Cluster and Individual Randomization

While designing a trial to evaluate a complex intervention, one may be confronted with the dilemma that randomization at the level of the individual patient risks contamination bias, whereas cluster randomization risks incomparability of study arms and recruitment problems. Literature provides only few solutions to this dilemma and these are not always feasible. As an alternative solution for this dilemma, we developed a new two-stage randomization method called pseudo cluster randomization. In the first stage, the clusters (e.g., recruiting physicians) are randomized into two groups: one group of clusters in which the majority of the participants (e.g., 80%) will receive the experimental treatment; one group of clusters in which the majority will receive the control condition. Following this, the second stage of the randomization involves randomly assigning participants within clusters in the proportions determined by the first stage. This has important advantages. Compared with cluster randomization the potential occurrence of baseline incomparability of treatment arms and poor recruitment is reduced, because the physicians who recruit the participants are unable to know in advance which treatment condition the next participant they recruit will be assigned to. Limiting the exposure of half of the physicians to the innovative intervention lowers risk of contamination bias. When this type of contamination bias is present, pseudo cluster randomization can be more efficient than individual or cluster randomization in that smaller number of study participants is needed to achieve a predefined power.

Improved Vitamin D Status Despite Allocation to Placebo in a Tuberculosis Treatment Trial: Contamination Bias?

Improved Vitamin D Status Despite Allocation to Placebo in a Tuberculosis Treatment Trial: Contamination Bias?

Improved Vitamin D Status Despite Allocation to Placebo in a Tuberculosis Treatment Trial: Contamination Bias?

Improved Vitamin D Status Despite Allocation to Placebo in a Tuberculosis Treatment Trial: Contamination Bias?

Contamination in trials of educational interventions

To consider the effects of contamination on the magnitude and statistical significance (or precision) of the estimated effect of an educational intervention, to investigate the mechanisms of contamination, and to consider how contamination can be avoided. Major electronic databases were searched up to May 2005. An exploratory literature search was conducted. The results of trials included in previous relevant systematic reviews were then analysed to see whether studies that avoided contamination resulted in larger effect estimates than those that did not. Experts' opinions were elicited about factors more or less likely to lead to contamination. We simulated contamination processes to compare contamination biases between cluster and individually randomised trials. Statistical adjustment was made for contamination using Complier Average Causal Effect analytic methods, using published and simulated data. The bias and power of cluster and individually randomised trials were compared, as were Complier Average Causal Effect, intention-to-treat and per protocol methods of analysis. Few relevant studies quantified contamination. Experts largely agreed on where contamination was more or less likely. Simulation of contamination processes showed that, with various combinations of timing, intensity and baseline dependence of contamination, cluster randomised trials might produce biases greater than or similar to those of individually randomised trials. Complier Average Causal Effect analyses produced results that were less biased than intention-to-treat or per protocol analyses. They also showed that individually randomised trials would in most situations be more powerful than cluster randomised trials despite contamination. The probability, nature and process of contamination should be considered when designing and analysing controlled trials of educational interventions in health. Cluster randomisation may or may not be appropriate and should not be uncritically assumed always to be a solution. Complier Average Causal Effect models are an appropriate way to adjust for contamination if it can be measured. When conducting such trials in future, it is a priority to report the extent, nature and effects of contamination.

Pseudo cluster randomization: a treatment allocation method to minimize contamination and selection bias

In some clinical trials, treatment allocation on a patient level is not feasible, and whole groups or clusters of patients are allocated to the same treatment. If, for example, a clinical trial is investigating the efficacy of various patient coaching methods and randomization is done on a patient level, then patients who are receiving different methods may come into contact with each other and influence each other. This would create contamination of the treatment effects. Such bias might be prevented by randomization on the coaches level. The patients of a coach constitute a cluster and all the subjects in that cluster receive the same treatment. Disadvantages of this approach may be reduced statistical efficiency and recruitment bias, as the treatment that a subject will receive is known in advance. Pseudo cluster randomization avoids this, because in pseudo cluster randomization, not everybody in a certain cluster receives the same treatment, just the majority. There are two groups of clusters: in one group the majority of subjects receive treatment A, while a limited number receive treatment B. In the other group of clusters the proportions are reversed. The statistical properties of this method are described. When contamination is present, the method appears to be more efficient than randomization on a patient level or on a cluster level.

Large-scale variations of the dust optical properties in the Galaxy

We present an analysis of the dust optical properties at large scale, for the whole galactic anticenter hemisphere. We used the 2MASS Extended Source Catalog to obtain the total reddening on each galaxy line of sight and we compared this value to the IRAS 100 microns surface brightness converted to extinction by Schlegel et al (1998). We performed a careful examination and correction of the possible systematic effects resulting from foreground star contamination, redshift contribution and galaxy selection bias. We also evaluated the contribution of dust temperature variations and interstellar clumpiness to our method. The correlation of the near-infrared extinction to the far-infrared optical depth shows a discrepancy for visual extinction greater than 1 mag with a ratio A_V(FIR) / A_V(gal) = 1.31 +- 0.06. We attribute this result to the presence of fluffy/composite grains characterized by an enhanced far--infrared emissivity. Our analysis, applied to half of the sky, provides new insights on the dust grains nature suggesting fluffy grains are found not only in some very specific regions but in all directions for which the visual extinction reaches about 1 mag.