Predicting the Direction of Causal Effect Based on an Instrumental Variable Analysis: A Cautionary Tale

Burgess, Stephen; Small, Dylan S.

doi:10.1515/jci-2015-0024

Cited by 15 publications

(13 citation statements)

References 33 publications

(37 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If we find that then, assuming Z is an instrument, we have no information about the direction or the size of an effect. This observation has been made for time-fixed treatments [ 9 ]; here we extend it to time-varying treatments as well.…”

Section: On Evidence Regarding the Direction And Magnitude Of Causal supporting

confidence: 59%

“…When the average causal null hypothesis holds, and are not guaranteed to be equal without additional conditions. One additional condition that would guarantee equality is a monotonic treatment effect: for a binary A , the treatment is either beneficial or harmful for all individuals in the study population (e.g., for all individuals i ) [ 9 ]. By the contrapositive, whenever our estimates of and are not equal, we have evidence that at least one of the following is true: the average causal null does not hold, the proposed instrument is not an instrument, or the treatment effect is not monotonic.…”

Section: Causal Null Hypotheses For Time-fixed Treatmentsmentioning

confidence: 99%

“…Here, we consider different versions of causal null hypotheses, describe conditions under which the instrument-outcome association can be used to test these hypotheses, and discuss how to conduct and interpret results from these tests. We begin by reviewing established results in the simple setting of time-fixed treatments [ 9 ], and then extend our discussion to more realistic settings with time-varying treatments.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Causal null hypotheses of sustained treatment strategies: What can be tested with an instrumental variable?

2018

View full text Add to dashboard Cite

Sometimes instrumental variable methods are used to test whether a causal effect is null rather than to estimate the magnitude of a causal effect. However, when instrumental variable methods are applied to time-varying exposures, as in many Mendelian randomization studies, it is unclear what causal null hypothesis is tested. Here, we consider different versions of causal null hypotheses for time-varying exposures, show that the instrumental variable conditions alone are insufficient to test some of them, and describe additional assumptions that can be made to test a wider range of causal null hypotheses, including both sharp and average causal null hypotheses. Implications for interpretation and reporting of instrumental variable results are discussed.Electronic supplementary materialThe online version of this article (10.1007/s10654-018-0396-6) contains supplementary material, which is available to authorized users.

show abstract

Section: On Evidence Regarding the Direction And Magnitude Of Causal supporting

confidence: 59%

Section: Causal Null Hypotheses For Time-fixed Treatmentsmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Causal null hypotheses of sustained treatment strategies: What can be tested with an instrumental variable?

2018

View full text Add to dashboard Cite

show abstract

“…Importantly for MR applications and in contrast with the ITT estimate in a partial compliance type setting, we note that the estimate of the G-Y association is generally not interpretable in terms of a causal effect. Neither does it permit inference about the magnitude nor sign of a causal effect (Didelez and Sheehan 2007b;Burgess and Small 2016;. It is purely a test for a causal effect and further assumptions must be made to obtain a point estimate of such an effect should it seem likely to be present.…”

Section: Testing For a Causal Effect By Testing For A G-y Associationmentioning

confidence: 99%

Epidemiology, genetic epidemiology and Mendelian randomisation: more need than ever to attend to detail

Sheehan

Didelez

2019

Hum Genet

View full text Add to dashboard Cite

In the current era, with increasing availability of results from genetic association studies, finding genetic instruments for inferring causality in observational epidemiology has become apparently simple. Mendelian randomisation (MR) analyses are hence growing in popularity and, in particular, methods that can incorporate multiple instruments are being rapidly developed for these applications. Such analyses have enormous potential, but they all rely on strong, different, and inherently untestable assumptions. These have to be clearly stated and carefully justified for every application in order to avoid conclusions that cannot be replicated. In this article, we review the instrumental variable assumptions and discuss the popular linear additive structural model. We advocate the use of tests for the null hypothesis of 'no causal effect' and calculation of the bounds for a causal effect, whenever possible, as these do not rely on parametric modelling assumptions. We clarify the difference between a randomised trial and an MR study and we comment on the importance of validating instruments, especially when considering them for joint use in an analysis. We urge researchers to stand by their convictions, if satisfied that the relevant assumptions hold, and to interpret their results causally since that is the only reason for performing an MR analysis in the first place.

show abstract

“…Causal inference and missing data problems have been extensively researched in recent decades in medical, social, and economical sciences (eg, other works). Consider a medical experiment with n subjects where each subject is assigned to either the treatment or the control group.…”

Section: Introductionmentioning

confidence: 99%

Using a monotone single‐index model to stabilize the propensity score in missing data problems and causal inference

Qin

et al. 2018

Statistics in Medicine

View full text Add to dashboard Cite

The augmented inverse weighting method is one of the most popular methods for estimating the mean of the response in causal inference and missing data problems. An important component of this method is the propensity score.Popular parametric models for the propensity score include the logistic, probit, and complementary log-log models. A common feature of these models is that the propensity score is a monotonic function of a linear combination of the explanatory variables. To avoid the need to choose a model, we model the propensity score via a semiparametric single-index model, in which the score is an unknown monotonic nondecreasing function of the given single index. Under this new model, the augmented inverse weighting estimator (AIWE) of the mean of the response is asymptotically linear, semiparametrically efficient, and more robust than existing estimators. Moreover, we have made a surprising observation. The inverse probability weighting and AIWEs based on a correctly specified parametric model may have worse performance than their counterparts based on a nonparametric model. A heuristic explanation of this phenomenon is provided. A real-data example is used to illustrate the proposed methods.KEYWORDS causal inference, empirical process, inverse weighting, missing data, pool adjacent violation algorithm, single-index model 1442

show abstract

Predicting the Direction of Causal Effect Based on an Instrumental Variable Analysis: A Cautionary Tale

Cited by 15 publications

References 33 publications

Causal null hypotheses of sustained treatment strategies: What can be tested with an instrumental variable?

Causal null hypotheses of sustained treatment strategies: What can be tested with an instrumental variable?

Epidemiology, genetic epidemiology and Mendelian randomisation: more need than ever to attend to detail

Using a monotone single‐index model to stabilize the propensity score in missing data problems and causal inference

Contact Info

Product

Resources

About