Biomedical subjects

Bob Zhong

Publications and source records attributed to Bob Zhong.

4 recordsLinked to original sources

PubMed2004

Assessing the agreement between two quantitative assays with repeated measurements.

We propose a statistical test in evaluating the equivalence (or agreement) between assay values x and y obtained using two methods in assay or instrument validation, where both x and y are measurements of an unobserved analyte z. Our method builds on the availability of repeated measurements from each sampled subject, which enables us to assess the agreement in terms of the conditional mean of the difference x - y given z as well as the conditional variance of x - y given z. With the help of repeated measurements and a large number of sampled subjects, our method does not require any distributional assumption on (x, y, z) or any model assumption on the conditional mean (or variance) of x - y. Furthermore, our test is designed so that when the null hypothesis is rejected, we can conclude that the two assay methods do not have practically meaningful difference with a high statistical assurance, which is an approach adopted in the assessment of bioequivalence between two drug products.

Data Interpretation, Statistical↗

PubMed2003

Last observation carry-forward and last observation analysis.

Drop-out often occurs in clinical trials with multiple visits and drop-out is often informative in the sense that the population of patients who dropped out is different from the population of patients who completed the study. To handle data with informative drop-out, an intention-to-treat analysis, which evaluates treatment effects over the population of all randomized patients with at least one post-treatment evaluation, is often required by the regulatory agencies. As a popular and simple intention-to-treat analysis, the last observation carry-forward (LOCF) analysis of variance (ANOVA) performs a statistical test for treatment effects by treating the last observation prior to drop-out as the observation from the last visit. Although discussions, examples and limited empirical results about the LOCF analysis can be found, its theoretical property is unclear. We find that the LOCF one-way ANOVA test is actually asymptotically valid (that is, its asymptotic size is equal to the nominal size) in the special but important case where only two treatments are compared and the two treatment groups have the same number of patients, regardless of whether drop-out is informative or not. In other cases, however, the asymptotic size of the LOCF test is different from the nominal size and is often too small when drop-out is informative, which results in a loss in power of detecting treatment effects, a disadvantage to drug companies. We propose an asymptotically valid test for comparing the global means over subpopulations, where each subpopulation contains patients dropping out after a particular visit. Some simulation results are presented to study the finite sample performance of the LOCF test and our proposed test.

Analysis of Variance↗

PubMed2003

Evaluating the agreement of two quantitative assays with repeated measurements.

A common task in assay validation is to show the agreement between an assay under investigation and a reference assay. Hence, in the hypothesis setup, we should choose nonagreement as the null hypothesis so that when the null hypothesis is rejected at 5% level of significance, we have a 95% statistical assurance to claim the agreement between two assays. In this paper, we propose a statistical test with nonagreement as the null hypothesis. The calculation of sample size is also given. Some simulation results are provided for illustration.

Biometry↗

PubMed2002

Evaluating qualitative assays using sensitivity and specificity.

Sensitivity and specificity are two important indices of performance of qualitative assays. Evaluating these indices usually requires one to identify the true disease state of each subject involved in a study. This implies that a perfect test, a "gold standard," is needed to test each subject. However, a gold standard test cannot always be performed on all subjects, whether because of cost or adverse effect on a subject's welfare. In these situations, a common practice is to apply both a currently used assay and an investigational assay to the same specimen. If the testing results are discordant, a gold standard test is applied. This approach has been criticized by many and, in fact, the statistics based on this approach usually overestimate sensitivity and specificity. This paper proposes two alternative methods to estimate sensitivity and specificity. Simulation results show that these methods perform better than the commonly used existing ones. This paper proposes new acceptance criteria and designs to specific topics for the evaluation of blood related assays as well. To evaluate a qualitative assay related to blood specimens, one must also perform studies of storage conditions, interfering substances, and other related factors, in order to establish the equivalency of the assay under standard and various other conditions. To conduct these studies, true negative blood donor specimens are used as a sample from a nondiseased population; and blood donor specimens with spiked analyte are used to represent a sample from a diseased population. Currently, the target-spiking ranges and sample sizes are determined subjectively. This paper presents new acceptance criteria on acceptable conditions and objective standards for selecting the target-spiking range and sample size.

Algorithms↗