PubMed Health⌕ Search
Home/Archive/PubMed

PubMed · 3406605

Confidence intervals for causal parameters.

J M Robins
View original source ↗Export metadata ↓

Abstract

Consider an unbiased follow-up study designed to investigate the causal effect of a dichotomous exposure on a dichotomous disease outcome. Under a deterministic outcome model, a standard '95 per cent binomial confidence interval' may fail to cover the causal parameter of interest at the nominal rate when we take the causal parameter to be a parameter associated with the observed study population (regardless of whether the observed study population was sampled from a larger superpopulation). I propose new interval estimators that, in this setting, improve upon the performance of the standard 'binomial confidence interval.'

Explore related subjects

BiometryEpidemiologic MethodsFollow-Up StudiesRandom AllocationRisk FactorsJournal ArticleResearch Support, Non-U.S. Gov'tResearch Support, U.S. Gov't, P.H.S.

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

J M Robins. 1988. Confidence intervals for causal parameters.. https://doi.org/10.1002/sim.4780070707

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed1996

[Conservative experimental groups in clinical comparative trials. Tables providing the minimum number of patients].

To calculate the minimum size of sample required for clinical trials it is first necessary to decide the expected difference between the groups, and then the acceptable risk of wrongly declaring a significant result (alpha) or wrongly declaring a non-significant result (beta). Once these three parameters are set, simple tables can show the necessary sample sizes. Such tables are presented for classical comparative studies, survival studies, and group sequential plans. Sequential plans often entail large savings in sample size compared with fixed group plans.

Biometry↗
PubMed1996

Characterisation of the possible effect on birthweight following frequent prenatal ultrasound examinations.

The objective of this study was to evaluate and characterise by study of newborn biometry a possible effect on birthweight which we observed previously in a randomised controlled trial of multiple prenatal ultrasound examinations. A total of 2743 women with single pregnancies had been allocated at random to either a protocol of ultrasound imaging and continuous wave Doppler studies at 18, 24, 28, 34 and 38 weeks gestation (intensive group), or to a protocol of a single imaging examination at 18 weeks and further imaging scans only as clinically indicated (regular group). When compared with those in the regular group, and adjusted for other confounding variables, normally formed babies of term gestational age in the intensive group tended to be shorter when measured at birth (P = 0.123) and on day 2-3 of age (P = 0.068). There were statistically insignificant reductions in the circumferences of the chest, abdomen and mid-arm; and in the skinfold thicknesses of the triceps, parascapular and subscapular regions. Principal component analysis showed a trend for a reduction for the skeletal component (P = 0.085) but not for the soft tissue component (P = 0.332). Comparison of the neonatal biometry in the two groups is not conclusive, but the differential effects on the various growth parameters suggest that if multiple scans do indeed restrict fetal growth, the mechanism is more likely to be an effect on bone growth rather than a reduction in nutrient supply from the placenta.

Biometry↗