Analysis of Incomplete Multivariate Data from Repeated Measurement Experiments

Crépeau, Hélène; Koziol, James A.; Reid, N.; Ys, Yuh

doi:10.2307/2530875

Cited by 31 publications

(12 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Cytokine concentrations of the various diet groups were compared using the multivariate rank test according to Koziol et al [23] and Crepeau et al [24]. This is a nonparametric test for the comparison of growth curves, which accounts for missing data at certain time points.…”

Section: Discussionmentioning

confidence: 99%

Pro‐ and anti‐inflammatory cytokines in healthy volunteers fed various doses of fish oil for 1 year

Blok

Deslypère

Demacker

et al. 1997

Eur J Clin Investigation

120

View full text Add to dashboard Cite

Abstract. Dietary supplementation with n-3 fatty acids from fish oil alleviates inflammation in various chronic inflammatory disease states. Reductions in the production of pro-inflammatory cytokines interleukin 1b (IL1b), tumour necrosis factor alpha (TNF-a), and interleukin 6 (IL-6) have been seen in humans after short-term n-3 fatty acid supplementation. We investigated longterm effects of dietary n-3 fatty acids on circulating cytokine concentrations and on ex vivo stimulated whole-blood production of IL-1b, TNF-a and interleukin 1 receptor antagonist (IL-1Ra), the naturally occurring antagonist of IL-1. A total of 58 monks with a mean age of 56 years were randomized into four groups and their diets were supplemented with 0, 3, 6, or 9 g of fish oil, providing 0, 1·06, 2·13 or 3·19 g of n-3 fatty acids per day. Subjects received equal amounts of saturated fatty acids, vitamin E and cholesterol. Compliance was excellent and erythrocyte fatty acid profiles closely reflected the amounts of n-3 fatty acids ingested. In the group receiving 9 g of fish oil per day, no influence of n-3 fatty acids on circulating cytokine concentrations was observed relative to placebo. Endotoxin-stimulated whole-blood cytokine production was measured at 26 and 52 weeks after the start and at 4, 8 and 26 weeks after cessation of supplementation. In all groups, the production of IL-1b and IL-1Ra was higher during supplementation than afterwards. However, no differences in cytokine production were noted between the placebo group and the various treatment groups at any point in time. Our results suggest that long-term supplementation of fish oil does not affect ex vivo cytokine production in man.

show abstract

Section: Discussionmentioning

confidence: 99%

Pro‐ and anti‐inflammatory cytokines in healthy volunteers fed various doses of fish oil for 1 year

Blok

Deslypère

Demacker

et al. 1997

Eur J Clin Investigation

120

View full text Add to dashboard Cite

show abstract

“…We will now consider a relatively small example with a continuous response, analyzed in Crépeau, Koziol, Reid and Yuh (1985). Fifty-four rats were divided into five treatment groups corresponding to exposure to increasing doses of halothane (0%, 0.25%, 0.5%, 1% and 2%).…”

Section: Examplesmentioning

confidence: 99%

Likelihood based frequentist inference when data are missing at random

Kenward

Molenberghs²

1998

Statist. Sci.

137

130

View full text Add to dashboard Cite

Abstract. One of the most often quoted results from the original work of Rubin and Little on the classification of missing value processes is the validity of likelihood based inferences under missing at random (MAR) mechanisms. Although the sense in which this result holds was precisely defined by Rubin, and explored by him in later work, it appears to be now used by some authors in a general and rather imprecise way, particularly with respect to the use of frequentist modes of inference. In this paper an exposition is given of likelihood based frequentist inference under an MAR mechanism that shows in particular which aspects of such inference cannot be separated from consideration of the missing value mechanism. The development is illustrated with three simple setups: a bivariate binary outcome, a bivariate Gaussian outcome and a two-stage sequential procedure with Gaussian outcome and with real longitudinal examples, involving both categorical and continuous outcomes. In particular, it is shown that the classical expected information matrix is biased and the use of the observed information matrix is recommended.

show abstract

“…When feasible, experimental protocols that assay the response to the levels of the stimulus in random order are to be preferred. Second, the analysis of response curve data will be immensely more complicated if the data are either unbalanced or incomplete (Crepeau et al 1985, Jennrich and Schluchter 1986, Davis and Wei 1988. In physiological ecology, imbalance is most likely to occur as uneven numbers of replicates within the experimental groups.…”

Section: Discussionmentioning

confidence: 99%