Nonlinear Predictive Models for Multiple Mediation Analysis: With an Application to Explore Ethnic Disparities in Anxiety and Depression Among Cancer Survivors

  • Qingzhao Yu
  • Kaelen L. Medeiros
  • Xiaocheng Wu
  • Roxanne E. Jensen


Mediation analysis allows the examination of effects of a third variable (mediator/confounder) in the causal pathway between an exposure and an outcome. The general multiple mediation analysis method (MMA), proposed by Yu et al., improves traditional methods (e.g., estimation of natural and controlled direct effects) to enable consideration of multiple mediators/confounders simultaneously and the use of linear and nonlinear predictive models for estimating mediation/confounding effects. Previous studies find that compared with non-Hispanic cancer survivors, Hispanic survivors are more likely to endure anxiety and depression after cancer diagnoses. In this paper, we applied MMA on MY-Health study to identify mediators/confounders and quantify the indirect effect of each identified mediator/confounder in explaining ethnic disparities in anxiety and depression among cancer survivors who enrolled in the study. We considered a number of socio-demographic variables, tumor characteristics, and treatment factors as potential mediators/confounders and found that most of the ethnic differences in anxiety or depression between Hispanic and non-Hispanic white cancer survivors were explained by younger diagnosis age, lower education level, lower proportions of employment, less likely of being born in the USA, less insurance, and less social support among Hispanic patients.


ethnic disparity health-related quality of life mediation/confounding analysis MY-Health study nonlinear models 



This study was partially funded by the NIMHD award # 1R15MD012387 and by the Louisiana State University Health Sciences Center Pilot Fund.

Compliance with Ethical Standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical Approval

This article does not contain any studies with human participants or animals performed by any of the authors.

Supplementary material

11336_2018_9612_MOESM1_ESM.pdf (172 kb)
Supplementary material 1 (pdf 171 KB)
11336_2018_9612_MOESM2_ESM.pdf (76 kb)
Supplementary material 2 (pdf 76 KB)


  1. Albert, J. M. (2008). Mediation analysis via potential outcomes models. Statistics in Medicine, 27(8), 1282–1304.CrossRefPubMedGoogle Scholar
  2. Alwin, D. F., & Hauser, R. M. (1975). The decomposition of effects in path analysis. American Sociological Review, 40(1), 37.CrossRefGoogle Scholar
  3. Avin, C., Shpitser, I., & Pearl, J. (2005). Identifiability of path-specific effects. In M. van Zaanen, T. Oates, G. Paliouras, & C. de la Higuera (Eds.), Proceedings of the international joint conference on artificial intelligence IJCAI-05. Edinburgh: Morgan-Kaufmann Publishers.Google Scholar
  4. 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(6), 1173–1182.CrossRefPubMedGoogle Scholar
  5. Breiman, L., Friedman, J., Olshen, R. A., & Stone, C. (1984). Classification and regression trees. Pacific Grove: Wadsworth.Google Scholar
  6. Culver, J. L., Arena, P. L., Antoni, M. H., & Carver, C. S. (2002). Coping and distress among women under treatment for early stage breast cancer: Comparing African Americans, Hispanics and non-Hispanic Whites. Psycho-Oncology, 11(6), 495–504.CrossRefPubMedGoogle Scholar
  7. Fan, Y. (2012). Multiple mediation analysis with general predictive models, Dissertation. New Orleans, LA: Louisiana State University Health Sciences Center.Google Scholar
  8. Friedman, J. (2001). Greedy function approximation: A gradient boosting machine. The Annals of Statistics, 29, 1189–1232.CrossRefGoogle Scholar
  9. Gallicchio, L., Calhoun, C., & Helzlsouer, K. J. (2013). Association between race and physical functioning limitations among breast cancer survivors. Supportive Care in Cancer, 22(4), 1081–1088.CrossRefPubMedGoogle Scholar
  10. Gu, C. (2013). Smoothing spline ANOVA models. New York: Springer.CrossRefGoogle Scholar
  11. Have, T. R. T., Joffe, M. M., Lynch, K. G., Brown, G. K., Maisto, S. A., & Beck, A. T. (2007). Causal mediation analyses with rank preserving models. Biometrics, 63(3), 926–934.CrossRefPubMedGoogle Scholar
  12. Jensen, R. E., Moinpour, C. M., Keegan, T. H. M., Cress, R. D., Wu, X.-C., & Paddock, L. (2017). The measuring your health study: Leveraging community-based cancer registry recruitment to establish a large, diverse cohort of cancer survivors for analyses of measurement equivalence and validity of the patient reported outcomes measurement information system (promis) short form items. Psychological Test and Assessment Modeling, 58(1), 99–117.Google Scholar
  13. Jensen, R. E., Moinpour, C. M., Potosky, A. L., Lobo, T., Hahn, E. A., Hays, R. D., et al. (2016). Responsiveness of 8 patient-reported outcomes measurement information system (PROMIS) measures in a large, community-based cancer study cohort. Cancer, 123(2), 327–335.CrossRefPubMedPubMedCentralGoogle Scholar
  14. Jensen, R. E., Potosky, A. L., Reeve, B. B., Hahn, E., Cella, D., Fries, J., et al. (2015). Validation of the PROMIS physical function measures in a diverse US population-based cohort of cancer patients. Quality of Life Research, 24(10), 2333–2344.CrossRefPubMedPubMedCentralGoogle Scholar
  15. Judd, C. M., & Kenny, D. A. (1981). Process analysis: Estimating mediation in treatment evaluations. Evaluation Review, 5(5), 602–619.CrossRefGoogle Scholar
  16. MacKinnon, D. P. (2000). Equivalence of the mediation, confounding and suppression effect. Prevention Science, 1(4), 173–181.CrossRefPubMedPubMedCentralGoogle Scholar
  17. Mackinnon, D. P. (2008). Introduction to statistical mediation analysis. New York: Lawrence Erlbaum Associates.Google Scholar
  18. Mackinnon, D. P., & Dwyer, J. H. (1993). Estimating mediated effects in prevention studies. Evaluation Review, 17(2), 144–158.CrossRefGoogle Scholar
  19. NIH. Calibration studies testing. Accessed 09 Feb 2017.
  20. NIH. Promis domain frameworks. Accessed 09 Feb 2017.
  21. NIH. Promis methodology. Accessed 09 Feb 2017.
  22. Patrick, D., & Erickson, P. (1993). Health policy, quality of life: Health care evaluation and resource allocation. New York, NY: Oxford University Press.Google Scholar
  23. Pearl, J. (2001). Direct and indirect effects. In Proceedings of the seventeenth conference on uncertainty and artificial intelligence. Morgan Kaufmann.Google Scholar
  24. Reeve, B. B., Potosky, A. L., Smith, A. W., Han, P. K., Hays, R. D., Davis, W. W., et al. (2009). Impact of cancer on health-related quality of life of older americans. JNCI Journal of the National Cancer Institute, 101(12), 860–868.CrossRefPubMedPubMedCentralGoogle Scholar
  25. Reeve, B. B., Stover, A. M., Jensen, R. E., Chen, R. C., Taylor, K. L., Clauser, S. B., et al. (2012). Impact of diagnosis and treatment of clinically localized prostate cancer on health-related quality of life for older Americans. Cancer, 118(22), 5679–5687.CrossRefPubMedPubMedCentralGoogle Scholar
  26. Reeve, B. B., & Terest, J. A. (2016). Overview to the two-part series: Measurement equivalence of the patient reported outcomes measurement information system (promis) short forms. Psychological Test and Assessment Modeling, 58(1), 31–35.Google Scholar
  27. Reeve, B. B., Thissen, D., DeWalt, D. A., Huang, I.-C., Liu, Y., Magnus, B., et al. (2015). Linkage between the PROMIS® pediatric and adult emotional distress measures. Quality of Life Research, 25(4), 823–833.CrossRefPubMedPubMedCentralGoogle Scholar
  28. Robins, J. M., & Greenland, S. (1992). Identifiability and exchangeability for direct and indirect effects. Epidemiology, 3(2), 143–155.CrossRefPubMedGoogle Scholar
  29. Rodenberg, C. A. (2006). A review of: Assessing quality of life in clinical trials: Theory and methods, second edition, by P. Fayers and R. Hays (eds.). Journal of Biopharmaceutical Statistics, 16(5), 761–763.CrossRefGoogle Scholar
  30. Teresi, J. A., Ocepek-Welikson, K., Kleinman, M., Ramirez, M., & Kim, G. (2016a). Measurement equivalence of the patient reported outcomes measurement information system (promis) anxiety short forms in ethnically diverse groups. Psychological Test and Assessment Modeling, 58(1), 183–219.PubMedPubMedCentralGoogle Scholar
  31. Teresi, J. A., Ocepek-Welikson, K., Kleinman, M., Ramirez, M., & Kim, G. (2016b). Psychometric properties and performance of the patient reported outcomes measurement information system (promis) depression short forms in ethnically diverse groups. Psychological Test and Assessment Modeling, 58(1), 141–181.PubMedPubMedCentralGoogle Scholar
  32. VanderWeele, T. J. (2009). Marginal structural models for the estimation of direct and indirect effects. Epidemiology, 20(1), 18–26.CrossRefPubMedGoogle Scholar
  33. VanderWeele, T. J., & Robinson, W. R. (2014). On the causal interpretation of race in regressions adjusting for confounding and mediating variables. Epidemiology, 25(4), 473–484.CrossRefPubMedPubMedCentralGoogle Scholar
  34. Vanderweele, T. J., & Vansteelandt, S. (2009). Conceptual issues concerning mediation, interventions and composition. Statistics and Its Interface, 2(4), 457–468.CrossRefGoogle Scholar
  35. VanderWeele, T., Vansteelandt, S., & Daniel, R. M. (2014). Effect decomposition in the presence of an exposure-induced mediator-outcome confounder. Epidemiology, 25(2), 300–306.CrossRefPubMedPubMedCentralGoogle Scholar
  36. Vansteelandt, S., & Daniel, R. M. (2017). Interventional effects for mediation analysis with multiple mediators. Epidemiology, 28(2), 258–265.CrossRefPubMedPubMedCentralGoogle Scholar
  37. Yost, K., Eton, D., Garcia, S., & Cella, D. (2011). Minimally important differences were estimated for six promis cancer scales in advanced-stage cancer patients. Journal of Clinical Epidemiology, 64(5), 507–516.CrossRefPubMedPubMedCentralGoogle Scholar
  38. Yu, Q., Fan, Y., & Wu, X. (2014). General multiple mediation analysis with an application to explore racial disparities in breast cancer survival. Journal of Biometrics & Biostatistics, 5(189), 189.Google Scholar
  39. Yu, Q., & Li, B. (2017). mma: An r package for multiple mediation analysis. Journal of Open Research Software, 5, 11.CrossRefGoogle Scholar
  40. Yu, Q., Li, B., & Scribner, R. A. (2009). Hierarchical additive modeling of nonlinear association with spatial correlations—An application to relate alcohol outlet density and neighborhood assault rates. Statistics in Medicine, 28(14), 1896–1912.CrossRefPubMedGoogle Scholar
  41. Yu, Q., Scribner, R., Leonardi, C., Zhang, L., Park, C., Chen, L., et al. (2017). Exploring racial disparity in obesity: A mediation analysis considering geo-coded environmental factors. Spatial and Spatio-temporal Epidemiology, 21, 13–23.CrossRefPubMedGoogle Scholar

Copyright information

© The Psychometric Society 2018

Authors and Affiliations

  1. 1.Biostatistics Program, School of Public HealthLouisiana State University Health Sciences CenterNew OrleansUSA
  2. 2.American College of SurgeonChicagoUSA
  3. 3.Louisiana Tumor RegistryNew OrleansUSA
  4. 4.Cancer Prevention and Control ProgramLombardi Comprehensive Cancer CenterWashingtonUSA

Personalised recommendations