Top

Quality of Life Research

Gepubliceerd in:

01-03-2009

Mediators, moderators, and modulators of causal effects in clinical trials––Dynamically Modified Outcomes (DYNAMO) in health-related quality of life

Auteurs: Gary W. Donaldson, Yoshio Nakamura, Carol Moinpour

Gepubliceerd in: Quality of Life Research | Uitgave 2/2009

Abstract

Purpose

There is more than one effect of a randomized intervention, and more than one response to a treatment. The mean group difference in the pre-specified outcome of a randomized controlled trial (RCT) estimates the average causal effect of treatment across causal mechanisms that may be distinct. To understand the comprehensive impact of an intervention, we propose identifying and separating the direct causal effects of the intervention from mediating, moderating, and modulating (individual differences) influences of uncontrolled variables.

Methods

Relational outcomes and moderated interventions describe two common mechanisms by which treatment interactions with uncontrolled variables expand or qualify the causal inferences to be drawn from an RCT, signifying treatment impact beyond that captured by conventional mean differences. Relational outcomes are associations between post-randomization measures. The treatment intervention may affect relational outcomes, and individual differences may modulate them. With moderated interventions, the effect of treatment on the outcome of interest depends on personal attributes or pre-randomization uncontrolled variables.

Results

Awareness and measurement of both types of mechanisms can greatly improve interpretation of the results of a clinical trial.

Conclusion

The integrated formal system of Dynamically Modified Outcomes (DYNAMO) provides comprehensive analysis of the diverse causal influences and interactions operating in a clinical trial.

volgende artikel Chemotherapeutic impact on pain and global health-related quality of life in hormone-refractory prostate cancer: Dynamically Modified Outcomes (DYNAMO) analysis of a randomized controlled trial

In the language of time series, pain and anxiety are stationary in that their overall levels and variances do not change over long time periods.

To simplify presentation of the concepts, we represent anxiety as having abrupt, immediate effects on pain, with no reciprocal causation. Considerations are similar, though more complex, when onset and offset are gradual, when pain also causes anxiety, and with varying degrees of random error in the relationships.

It is of course highly desirable to reduce levels of pain and anxiety over time, so that the lines would trend downward, but the difficulty in doing this is precisely what makes these conditions chronic. Clinical management therefore emphasizes control and functioning, intrinsically relational concepts, in addition to reduction of intensity.

For simplicity of presentation, we assume both X and W are binary variables coded 0 or 1, but the model generalizes naturally to polytomous and continuous variables.

We retain this orientation for hypothetical manipulations throughout without further comment. Completely equivalent, though less intuitive, are hypothetical manipulations that replace the treatment intervention with the control intervention.

For reference, the regression line of Y on Z for Controls [E(Y|Z, X = 0)] is indicated; this line is relevant for individual mediating and causal effects but does not figure in the calculation of the ACE.

Intercept terms, which change with the addition or modification of explanatory variables, can easily obscure the conceptual meanings presented in Table 1. It is possible to incorporate correct adjustments for intercepts, but slopes and mean levels have more natural interpretations in this context. To estimate the parameters of Table 1 with standard software, it is most efficient to combine slope estimates (only) from regression modules with descriptive estimates of population means.

Literally, “contrary to fact,” yet the literal meaning fails to convey the scope of the concept. Counterfactual reasoning is at the heart of scientific explanation, at two levels [1]. Scientific laws express general hypothetical relationships describing what would happen were certain conditions to be modified (given that they are not so modified now), and specific counterfactual reasoning allows us to predict what would have happened had a particular outcome been otherwise than we observe.

Pearl, J. (2000). Causality: Models, reasoning, and inference. Cambridge: Cambridge University Press.

Gatchel, R. J., Peng, Y. B., Peters, M. L., Fuchs, P. N., & Turk, D. C. (2007). The biopsychosocial approach to chronic pain: Scientific advances and future directions. Psychological Bulletin, 133, 581–624. doi:10.1037/0033-2909.133.4.581.CrossRefPubMed

Keogh, E., & Asmundson, G. J. (2004). Negative affectivity catastrophizing and anxiety sensitivity. In G. J. Asmundson, J. W. S. Vlaeyen, & G. Crombez (Eds.), Understanding and treating fear of pain (pp. 91–115). Oxford: Oxford University Press.

Keogh, E., & Birkby, J. (1999). The effect of anxiety sensitivity and gender on the experience of pain. Cognition and Emotion, 13, 813–829. doi:10.1080/026999399379096.CrossRef

Keogh, E., & Cochrane, M. (2002). Anxiety sensitivity, cognitive biases, and the experience of pain. Journal of Pain, 3, 320–329. doi:10.1054/jpai.2002.125182.CrossRefPubMed

Keogh, E., Hamid, R., Hamid, S., & Ellery, D. (2004). Investigating the effect of anxiety sensitivity, gender and negative interpretive bias on the perception of chest pain. Pain, 111, 209–217. doi:10.1016/j.pain.2004.06.017.CrossRefPubMed

Lindley, D. (2002). Seeing and doing: The concept of causation. International Statistical Review. Revue Internationale de Statistique, 70, 191–214.CrossRef

Pearl, J. (1995). Causal diagrams for empirical research. Biometrika, 82, 669–710. doi:10.1093/biomet/82.4.669.CrossRef

Steyer, R. (2005). Analyzing individual and average causal effects via structural equation models. Methodology, 1, 39–54.

10.

Spirtes, P. G. C., & Sheines, R. (1993). Causation, prediction, and search (2nd ed.). New York: MIT Press.

11.

Baron, R., & Kenny, D. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. doi:10.1037/0022-3514.51.6.1173.CrossRefPubMed

12.

Judd, C., & Kenny, D. (1981). Process analysis: Estimating direction in evaluation research. Evaluation Research, 9, 602–618. doi:10.1177/0193841X8100500502.CrossRef

13.

MacKinnon, D., Lockwood, C., Hoffman, J., West, S., & Sheets, V. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7, 83–104. doi:10.1037/1082-989X.7.1.83.CrossRefPubMed

14.

Albert, J. M. (2008). Mediation analysis via potential outcomes models. Statistics in Medicine, 27(8), 1282–1304. doi:10.1002/sim.3016.CrossRefPubMed

15.

Kraemer, H., Wilson, G., Fairburn, C., & Agras, W. (2002). Mediators and moderators of treatment effects in randomized clinical trials. Archives of General Psychiatry, 59, 877–883. doi:10.1001/archpsyc.59.10.877.CrossRefPubMed

16.

Kraemer, H. C., Lowe, K. K., & Kupfer, D. J. (2005). To your health: What research tells us about risk. New York: Oxford University Press.

17.

Kraemer, H. C., Stice, E., Kazdin, A., Offord, D., & Kupfer, D. (2001). How do risk factors work together? mediators, moderators, and independent, overlapping, and proxy risk factors. The American Journal of Psychiatry, 158, 848–856.PubMed

18.

Bellamy, S. L., Lin, J. Y., & Ten Have, T. R. (2007). An introduction to causal modeling in clinical trials. Clinical Trials, 4, 58–73. doi:10.1177/1740774506075549.CrossRefPubMed

19.

Kenny, D. A. (2008). Reflections on mediation. Organizational Research Methods, 11, 353–358. doi:10.1177/1094428107308978.CrossRef

20.

Lauritzen, S. G. (1996). Graphical models. Oxford: Clarendon Press.

21.

Muthen, L. & Muthen, B. (1998–2005). Mplus user’s guide. Los Angeles: Muthen & Muthen.

22.

Rosenbaum, P. R. (1984). From association to causation in observational studies. the role of tests of strongly ignorable treatment assignment. Journal of the American Statistical Association, 79, 41–48. doi:10.2307/2288332.CrossRef

23.

Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41–55. doi:10.1093/biomet/70.1.41.CrossRef

24.

Rosenbaum, P. R., & Rubin, D. B. (1984). Reducing bias in observational studies using subclassification on the propensity score. Journal of the American Statistical Association, 79, 516–524. doi:10.2307/2288398.CrossRef

25.

Rubin, D. B. (1974). Estimating causal effects of treatments in randomized and nonrandomized studies. Journal of Educational Psychology, 66, 688–701. doi:10.1037/h0037350.CrossRef

26.

Rubin, D. B. (1978). Bayesian inference for causal effects. The role of randomization. Annals of Statistics, 6, 34–58. doi:10.1214/aos/1176344064.CrossRef

27.

Rubin, D. B. (1986). Which ifs have causal answers. Journal of the American Statistical Association, 81, 961–962. doi:10.2307/2289065.CrossRef

28.

Stone, (1993). The assumptions on which causal inferences rest. Journal of the Royal Statistical Society, B, 55, 455–466.

29.

Jo, B. (2002). Model misspecification sensitivity analysis in estimating causal effects of interventions with noncompliance. Statistics in Medicine, 21, 3161–3181. doi:10.1002/sim.1267.CrossRefPubMed

30.

Pearl, J., & Verma, T. (1991). A theory of inferred causation. In: J. A. Allen, R. Fikes & E. Sandewell (Eds.), Principles of knowledge representation and reasoning: Proc. 2nd Int. Conf., (pp.441–452). San Mateo, CA: Morgan Kaufmann.

31.

Moinpour, C. M., Donaldson, G. W., & Redman, M. W. (2007). Do general dimensions of quality of life add value to symptoms data? Journal of the National Cancer Institute Monographs, 37, 31–38. doi:10.1093/jncimonographs/lgm007.CrossRefPubMed

32.

Scheines, R., Spirtes, P., Glymour, C., & Meek, C. (2000). Tetrad ii: Tools for discovery. Hillsdale, NJ: Erlbaum.

33.

Edwards, D. (2000). Introduction to graphical modeling (2nd ed.). New York: Springer.

34.

Edwards, D. (2004). Mim 3.2. Free Software Foundation: Boston.

35.

Box, G. E. P. (1949). A general distribution theory for a class of likelihood criteria. Biometrika, 36, 317–346.PubMed

Titel: Mediators, moderators, and modulators of causal effects in clinical trials––Dynamically Modified Outcomes (DYNAMO) in health-related quality of life
Auteurs: Gary W. Donaldson
Yoshio Nakamura
Carol Moinpour
Publicatiedatum: 01-03-2009
Uitgeverij: Springer Netherlands
Gepubliceerd in: Quality of Life Research / Uitgave 2/2009
Print ISSN: 0962-9343
Elektronisch ISSN: 1573-2649
DOI: https://doi.org/10.1007/s11136-008-9439-x

Bohn Stafleu van Loghum

Deel dit onderdeel of sectie (kopieer de link)

Abstract

Purpose

Methods

Results

Conclusion

Log in om toegang te krijgen

Andere artikelen Uitgave 2/2009

Test–retest reliability of the Food Allergy Quality of Life Questionnaires (FAQLQ) for children, adolescents and adults

The first stage of developing preference-based measures: constructing a health-state classification using Rasch analysis

Health-related quality of life after coronary artery bypass grafting. The impact of a randomised controlled home-based intervention program

Obesity and health related quality of life in the general adult population of the Canary Islands

Does age modify the relationship between morbidity severity and physical health in English and Dutch family practice populations?

Exploring household income as a predictor of psychological well-being among long-term colorectal cancer survivors