Kernel Density Estimation for Random-effects Meta-analysis

Branko Miladinovic, Ambuj Kumar, Benjamin Djulbegovic

Abstract


Random-effects meta-analyses are commonly performed to combine estimates of treatment effect from
different studies in the presence of heterogeneity. The method incorporates between-study variance in the overall estimate
of summary effect and its standard error. In addition to calculating the summary effect which relates to the average
treatment effect across all trials, prediction intervals have been recommended to give a range for the predicted parameter
in a new study. As both the calculation of summary effects and prediction intervals rely on the assumption that the effects
underlying different studies are normally distributed, in this manuscript we demonstrate how distribution assumptionfree
weighted kernel density estimation can be used to construct a probability distribution of observed effect sizes, thus
gaining insight into the variability of summary effects. In our study, the weighted kernel density estimates are calculated
using the Gaussian kernel and the adaptive bandwidth selection process. The weights are incorporated based on five
different methods for estimating between-trial heterogeneity and sampling errors from all studies.

Full Text:

PDF

References


D. G. Altman, K. F. Schulz, D. Moher, et al., The revised CONSORT statement for reporting randomized trials:

explanation and elaboration, Ann Internal Medicine, 134 (2001) 663-694.

D. Bohning. Meta-analysis: a unifying meta-likelihood approach framing unobserved heterogeneity, study covariates,

publication bias, and study quality, Methods Inf Med, 44 (2005) 127-135.

W. G. Cochrane, The combination of estimates from different experiments, Biometrics, 10 (1954) 101-129.

D. Burr and H. Doss, A Bayesian semi-parametric model for random-effects meta-analysis, J Am Stat Assoc, 100

(2005) 242-251.

R. DerSimonian and N. Laird, Meta-analysis in clinical trials, Control Clin Trials, 7 (1986) 177-188.

R. DerSimonian and R. Kacker, Random-effects model for meta-analysis of clinical trials: an update, Contemp Clin

Trials, 28 (2007) 105-114.

B. Djulbegovic, A. Kumar, H. P. Soares, et al, Treatment success in cancer: new cancer treatment successes identified

in phase III randomized controlled trials conducted by the National Cancer Institute-sponsored cooperative

oncology groups, 1955 to 2006, Arch Internal Med, 168 (2008) 632-642.

J. P. Higgins, S. G. Thompson, and D. J. Spiegehalter, A re-evaluation of random-effects meta-analysis, J Roy Stat

Soc Ser A, 172 (2009) 137-159.

J. P. Higgins and S. Green, Cochrane handbook for systematic reviews of interventions, JohnWiley and Sons, 2008.

R. N. Kacker and D. A. Harville, Approximations for standard errors of estimators of fixed and random effects in

mixed linear models, J Am Stat Assoc, 79 (1984), 853-862.

K. J. Lee and S. G. Thompson, Flexible parametric models for random-effects distributions, Stat Med, 27 (2008)

-434.

D. I. Ohlssen, L. D. Sharples, and D. J. Spiegehalter, Flexible random-effects models using Bayesian semiparametric

models: applications to institutional comparisons, Stat Med, 26 (2007) 2088-2112.

R. C. Paule andJ. Mandel, Consensus values and weighting factors, J Res National Institute of Standards and

Technology, 87 (1982) 377-385.

M. Rosenblatt, Remarks on some nonparametric estimates of a density function, Ann Math Stat, 27 (1956) 832-837.

B. W. Silverman, Density estimation for statistics and data analysis, Chapman and Hall, London, 1986.

B.W. Silverman, Kernel density estimation using the fast Fourier transform, J Roy Stat Soc Ser C, 31 (1982) 93-99.

H. C. Sox, Defining comparative effectiveness research: the importance of getting it right, Med Care, 48 (2010)

S7-8.

T. D. Wager, M. Lindquist, and L. Kaplan, Meta-analysis of functional neuroimaging data; current and future

directions, Soc Cogn Affect Neurosci, 2 (2007), 150-158.

T. D. Wager, K. L. Phan, I. Liberzon, and S. F. Taylor, Valence, gender, and lateralization of functional brain

anatomy in emotion: a meta-analysis of findings from neuroimaging, Neuroimage, 19 (2003) 513-531.

K. J.Wells, J. S. Luque, B. Miladinovic, et al., Do community health worker interventions improve rates of screeing

mammography in the United States? A systematic review, Cancer Epidemiol Biomarkers Prev, 20 (2011) 1580-1598.


Refbacks

  • There are currently no refbacks.