Conclusions. The sensitivity and positive predictive value are generally more concerned than the specificity and negative predictive value in assessing and adjusting publication bias in meta-analyses. In this sense, Egger's regression can be recommended for its high sensitivity, while any positive result from Tang's method would suggest a probability of bias that should be taken seriously. Given the different patterns of the accuracy with the OR and the P1-P2 combination, a combination of Egger's regression and Tang's regression would be advisable. Further studies are needed to study the accuracy of methods used in combination. / Due to sampling error and true heterogeneity, a single study cannot provide a comprehensive picture and a precise estimate of, say the effectiveness of a treatment. Systematic reviews that identify and integrate relevant studies have become the most important scientific, quantitative method to summarize scientific research. Meta-analysis is the statistical method used in systematic reviews to combine results from individual studies. / However, due to selective submission and publication, not all relevant studies conducted, especially those unpublished studies with an insignificant negative result, are easily accessible to those who conduct reviews. As a result, the truth, say, the effect of a treatment, would be overestimated. This phenomenon is known as publication bias. A few methods for detecting the bias have been developed and used in meta-analyses. Although their accuracy has been studied, some important issues remain to be answered, such as when would a method be good enough for practical use and is it similarly good for different definitions of the odds ratio? / Methods. We conducted a simulation study to examine the accuracy of four commonly used bias-detection methods with various ORs and P1-P2 combinations. In a simulation study, the true bias status can be predetermined and thus be compared with the results of the bias-detection methods. The four methods are Egger's regression, funnel plot regression, rank correlation regression, and Tang's regression. Realistic sample size was used for simulating individual studies and the numbers of studies in a meta-analysis was also varied. Both the sensitivity and specificity are examined against the magnitude of the OR and the P1-P 2 combination to identify the ORs and P1-P 2 combinations for which a method is sufficiently accurate. Predictive values are also examined for the same reason and in the same manner. / Results. The sensitivity and positive predictive value are generally low and in particular when the OR is close to one for which publication bias is of a particular concern. Egger's regression has the highest sensitivity among the four, in particular when the OR is neither close to one nor exceptionally large or small. Due to the relatively lower specificity, the positive predictive value of Egger's regression is not as high as that for Tang's regression and funnel plot regression. Tang's regression and funnel plot regression are very similar in sensitivity, specificity and predictive values, with the former being slightly better. Rank correlation seems the least accurate method overall. Tang's regression has in general the highest positive predictive value among the four methods in particular when the OR is below one. / Chung Chi-keung. / "June 2006." / Adviser: Tang Jin Ling. / Source: Dissertation Abstracts International, Volume: 68-03, Section: B, page: 1588. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references (p. 116-124). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstracts in English and Chinese. / School code: 1307.

A clinical trial may have no subject (0%) or every subject (100%) developing the outcome of concern in either of the two comparison groups. This will cause a zero-cell in the four-cell (2x2) table of a trial using a binary outcome and make it impossible to estimate the odds ratio, a commonly used effect measure. A usual way to deal with this problem is to add 0.5 to each of the four cells in the 2x2 table. This is known as Haldane's approximation. In meta-analysis, Haldane's approximation can also be applied. Two approaches are possible: add 0.5 to only the trials with a zero cell or to all the trials in the meta-analysis. Little is known which approach is better when used in combination with different definitions of the odds ratio: the ordinary odds ratio, Peto's odds ratio and Mantel-Haenszel odds ratio. / A new formula is derived for converting Peto's odds ratio to the risk difference. The derived risk difference through the new method was then compared with the true risk difference and the risk difference derived by taking the Peto's odds ratio as the ordinary odds ratio. All simulations and analyses were conducted on the Statistical Analysis Software (SAS). / Conclusions. The estimated confidence interval of a meta-analysis would mostly exclude the truth if an inappropriate correction method is used to deal with zero cells. Counter-intuitively, the combined result of a meta-analysis will be worse as the number of studies included becomes larger. Mantel-Haenszel odds ratio without applying Haldane's approximation is recommended in general for dealing with sparse data in meta-analysis. The ordinary odds ratio with adding 0.5 to only the trials with a zero cell can be used when the trials are heterogeneous and the odds ratio is close to 1. Applying Haldane's approximation to all trials in a meta-analysis should always be avoided. Peto's odds ratio without Haldane's approximation can always be considered but the new formula should be used for converting Peto's odds ratio to the risk difference. / In addition, the odds ratio needs to be converted to a risk difference to aid decision making. Peto's odds ratio is preferable in some situations and the risk difference is derived by taking Peto's odds ratio as an ordinary odds ratio. It is unclear whether this is appropriate. / Methods. For studying the validity of Haldane's approximation, we defined 361 types of meta-analysis. Each type of meta-analysis is determined by a unique combination of the risk in the two compared groups and thus provides a unique true odds ratio. The number of trials in a meta-analysis is set at 5, 10 and 50 and the sample size of each trial in a meta-analysis varies at random but is made sufficiently small so that at least one trial in a meta-analysis will have a zero-cell. The number of outcome events in a comparison group of a trial is generated at random according to the pre-determined risk for that group. One thousand homogeneous meta-analyses and one thousand heterogeneous meta-analyses are simulated for each type of meta-analysis. Two Haldane's approximation approaches in addition to no approximation are evaluated for three definitions of the odds ratio. Thus, nine combined odds ratios are estimated for each type of meta-analysis and are all compared with the true odds ratio. The percentage of meta-analyses with the 95% confidence interval including the true odds ratio is estimated as the main index for validity of the correction methods. / Objectives. (1) We conducted a simulation study to examine the validity of Haldane's approximation as applied to meta-analysis, and (2) we derived and evaluated a new method to covert Peto's odds ratio to the risk difference, and compared it with the conventional conversion method. / Results. By using the true ordinary odds ratio, the percentage of meta-analyses with the confidence interval containing the truth was lowest (from 23.2% to 53.6%) when Haldane's approximation was applied to all the trials regardless the definition of the odds ratios used. The percentage was highest with Mantel-Haenszel odds ratio (95.0%) with no approximation applied. The validity of the corrections methods increases as the true odds ratio gets close to one, as the number of trials in a meta-analysis decreases, as the heterogeneity decreases and the trial size increases. / The proposed new formula performed better than the conventional method. The mean relative difference between the true risk difference and the risk difference obtained from the new formula is -0.006% while the mean relative difference between the true risk difference and the risk difference obtained from the conventional method is -10.9%. / The validity is relatively close (varying from 86.8% to 95.8%) when the true odds ratio is between 1/3 and 3 for all combinations of the correction methods and definitions of the odds ratio. However, Peto's odds ratio performed consistently best if the true Peto's odds ratio is used as the truth for comparison among the three definitions of the odds ratio regardless the correction method (varying from 88% to 98.7%). / Tam Wai-san Wilson. / "Jan 2006." / Adviser: J. L. Tang. / Source: Dissertation Abstracts International, Volume: 67-11, Section: B, page: 6488. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references (p. 151-157). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstracts in English and Chinese. / School code: 1307.

Interactions between cyanobacteria and herbivorous grazers play an important role in mediating the responses of freshwater phytoplankton assemblages to nutrient enrichment and top-down manipulation. Negative consequences associated with these interactions include dangerous blooms of harmful blue-green algae that have been implicated in the sickness and death of fishes, livestock, and, in extreme cases, humans. Frequently cited mechanisms influencing the interactions between grazers and cyanobacteria include cyanobacterial toxicity and morphology. To tease apart the importance of these mechanisms, I used meta-analysis to quantitatively synthesize the available literature on this topic. In addition, I conducted several experiments using novel techniques to determine the effect that cyanobacterial secondary metabolites from the bloom-forming cyanobacterium,

Vitamin K has a negligible effect on bone mineral density (BMD) and a large but uncertain effect on fractures. The three studies in the thesis explored uncertainty about the effect of vitamin K on fractures using the methods of economic evaluation and Bayesian meta-analysis. In study 1, a Markov probabilistic microsimulation model was developed for a hypothetical cohort of 50-year-old postmenopausal women without osteoporosis. This was a fracture incidence-based model, populated with data from the literature. It was used to examine the cost-effectiveness of two supplementation strategies over a lifetime horizon. We compared vitamin K2 (or vitamin K1) concurrent with vitamin D3 and calcium versus vitamin D3 and calcium alone. Study 2 included a systematic review, and classical and Bayesian univariate meta-analyses to determine the efficacies of the K vitamins on BMD or fractures in current and future trials. Study 3 used Bayesian bivariate random-effects meta-analysis to jointly model the treatment effects on two correlated bone outcomes. We compared the estimates from the univariate and bivariate meta-analyses and explored how these results would change the conclusions of the cost-effectiveness analysis. The strategies including vitamin K were highly cost-effective at willingness-to-pay of $50,000/QALY (quality-adjusted life year); however, the results were most sensitive to changes in the efficacy of vitamin K. The univariate meta-analyses showed large uncertainties in the anti-fracture effects of vitamin K2 in current and future trials. The bivariate 95% credible intervals were considerably narrower than those from the univariate meta-analyses. Using future odds ratios from the bivariate meta-analyses, vitamin K2 cost more than $100,000/QALY while vitamin K1 was cost-saving. Our analyses found substantial uncertainty around the estimates of the vitamin K effect on fractures. We recommend against routine use of vitamin K for fracture prevention. Bayesian bivariate meta-analysis accounts for all available information and should be considered when the treatment effects are measured on two correlated outcomes.

Vitamin K has a negligible effect on bone mineral density (BMD) and a large but uncertain effect on fractures. The three studies in the thesis explored uncertainty about the effect of vitamin K on fractures using the methods of economic evaluation and Bayesian meta-analysis. In study 1, a Markov probabilistic microsimulation model was developed for a hypothetical cohort of 50-year-old postmenopausal women without osteoporosis. This was a fracture incidence-based model, populated with data from the literature. It was used to examine the cost-effectiveness of two supplementation strategies over a lifetime horizon. We compared vitamin K2 (or vitamin K1) concurrent with vitamin D3 and calcium versus vitamin D3 and calcium alone. Study 2 included a systematic review, and classical and Bayesian univariate meta-analyses to determine the efficacies of the K vitamins on BMD or fractures in current and future trials. Study 3 used Bayesian bivariate random-effects meta-analysis to jointly model the treatment effects on two correlated bone outcomes. We compared the estimates from the univariate and bivariate meta-analyses and explored how these results would change the conclusions of the cost-effectiveness analysis. The strategies including vitamin K were highly cost-effective at willingness-to-pay of $50,000/QALY (quality-adjusted life year); however, the results were most sensitive to changes in the efficacy of vitamin K. The univariate meta-analyses showed large uncertainties in the anti-fracture effects of vitamin K2 in current and future trials. The bivariate 95% credible intervals were considerably narrower than those from the univariate meta-analyses. Using future odds ratios from the bivariate meta-analyses, vitamin K2 cost more than $100,000/QALY while vitamin K1 was cost-saving. Our analyses found substantial uncertainty around the estimates of the vitamin K effect on fractures. We recommend against routine use of vitamin K for fracture prevention. Bayesian bivariate meta-analysis accounts for all available information and should be considered when the treatment effects are measured on two correlated outcomes.

Thesis (D.P.H.)--University of Michigan.

Dissertation (D.P.H.)--University of Michigan.

Meta-analyses are conducted to synthesize the quantitative results of related studies. The random-effects meta-analysis model is based on the assumption that a distribution of true effects exists in the population. This distribution is often assumed to be normal with a mean and variance. The population variance, also called heterogeneity, can be estimated numerous ways. Accurate estimation of heterogeneity is necessary as a description of the distribution and for determining weights applied in the estimation of the summary effect when using inverse-variance weighting. To evaluate a wide range of estimators, we compared 16 estimators (Bayesian and non-Bayesian) of heterogeneity with regard to bias and mean square error over conditions based on reviews of educational and psychological meta-analyses. Three simulation conditions were varied: (a) sample size per meta-analysis, (b) true heterogeneity, and (c) sample size per effect size within each meta-analysis. Confidence or highest density intervals can be calculated for heterogeneity. The heterogeneity estimators that performed best over the widest range of conditions were paired with heterogeneity interval estimators. Interval estimators were evaluated based on coverage probability, interval width, and coverage of the estimated value. The combination of the Paule Manel estimator and Q-Profile interval method is recommended when synthesizing standardized mean difference effect sizes.

Reliability generalization (RG) is a method for meta-analysis of reliability coefficients to estimate average score reliability across studies, determine variation in reliability, and identify study-level moderator variables influencing score reliability. A total of 107 peer-reviewed RG studies published from 1998 to 2013 were systematically reviewed to characterize the meta-analytic methods employed and to evaluate quality of reporting practice against standards for transparency in meta-analysis reporting. Most commonly, RG studies meta-analyzed alpha coefficients, which were synthesized using an unweighted, fixed-effects model applied to untransformed coefficients. Moderator analyses most frequently included multiple regression and bivariate correlations employing a fixed-effects model on untransformed, unweighted coefficients. Based on a unit-weighted scoring system, mean reporting quality for RG studies was statistically less than that for a comparison study of 198 meta-analyses in the organizational sciences across 42 indicators; however, means were not statistically significantly different between the two studies when evaluating reporting quality on 18 indicators deemed essential to ethical reporting practice in meta-analyses. Since its inception a wide variety of statistical methods have been applied to RG, and meta-analysis of reliability coefficients has extended to fields outside of psychological measurement, such as medicine and business. A set of guidelines for conducting and reporting RG studies is provided.

Interferon gamma (IFN-γ) regulation of macrophages plays an essential role in innate immunity and pathogenicity of viral infections by directing large and small genome-wide changes in the transcriptional program of macrophages. Smaller changes at the transcriptional level are difficult to detect but can have profound biological effects, motivating the hypothesis of this thesis that responses of macrophages to immune activation by IFN-γ include small quantitative changes that are masked by noise but represent meaningful transcriptional systems in pathways against infection. To test this hypothesis, statistical meta-analysis of microarray studies is investigated as a tool to obtain the necessary increase in analysis sensitivity. Three meta-analysis models (Effect size model, Rank Product model, Fisher’s sum of logs) and three further modified versions were applied to a heterogeneous set of four microarray studies on the effect of IFN-γ on murine macrophages. Performance assessments include recovery of known biology and are followed by development of novel biological hypotheses through secondary analysis of meta-analysis outcomes in context of independent biological data sources. A separate network analysis of a microarray time course study investigate s if gene sets with coordinated time-dependent relationships overlap can also identify subtle IFN-γ related transcriptional changes in macrophages that match those identified through meta-analysis. It was found that all meta-analysis models can identify biologically meaningful transcription at enhanced sensitivity levels, with slightly improved performance advantages for a non-parametric model (Rank Product meta-analysis). Meta-analysis yielded consistently regulated genes, hidden in individual microarray studies, related to sterol biosynthesis (Stard3, Pgrmc1, Galnt6, Rab11a, Golga4, Lrp10), implicated in cross-talk between type II and type I interferon or IL-10 signalling (Tbk1, Ikbke, Clic4, Ptpre, Batf), and circadian rhythm (Csnk1e). Further network analysis confirms that meta-analysis findings are highly concentrated in a distinct immune response cluster of co-expressed genes, and also identifies global expression modularisation in IFN-γ treated macrophages, pointing to Trafd1 as a central anti-correlated node topologically linked to interactions with down-regulated sterol biosynthesis pathway members. Outcomes from this thesis suggest that small transcriptional changes in IFN-γ activated macrophages can be detected by enhancing sensitivity through combination of multiple microarray studies. Together with use of bioinformatical resources, independent data sets and network analysis, further validation assigns a potential role for low or variable transcription genes in linking type II interferon signalling to type I and TLR signalling, as well as the sterol metabolic network.