An Integrative Metaregression Framework for Descriptive Epidemiology

To provide the tools and knowledge needed in efforts to improve the health of the world’s populations, researchers collaborated on the Global Burden of Diseases, Injuries, and Risk Factors Study 2010. The study produced comprehensive estimates of more than 200 diseases and health risk factors in 187 countries over two decades, results that will be used by governments and non-governmental agencies to inform priorities for global health research, policies, and funding.

An Integrative Metaregression Framework for Descriptive Epidemiology is the first book-length treatment of model-based meta-analytic methods for descriptive epidemiology used in the Global Burden of Disease Study 2010. In addition to collecting the prior work on compartmental modeling of disease, this book significantly extends the model by formally connecting the system dynamics model of disease progression to a statistical model of epidemiological rates and demonstrates how the two models were combined to allow researchers to integrate all available relevant data. Practical applications of the model to meta-analysis of several different diseases complement the theoretical foundations of what the editors call the integrative systems modeling of disease in populations. The book concludes with a detailed description of the future directions for research in model-based meta-analysis of descriptive epidemiological data.

Abraham D. Flaxman is assistant professor of global health at the Institute for Health Metrics and Evaluation at the University of Washington. Theo Vos is professor of global health at the Institute for Health Metrics and Evaluation at the University of Washington. Christopher J. L. Murray is professor of global health and director of the Institute for Health Metrics and Evaluation at the University of Washington.

Figures

Tables

Acknowledgments

Introduction | by Abraham D. Flaxman, Theo Vos, and Christopher J.L. Murray

1. An Introductory Example

2. A Motivating Example

3. From Systematic Review to Metaregression

4. History of Generic Disease Modeling

5. What is Not in This Book

Section One | Theory and Methods

1. Background material on Bayesian methods / Abraham D. Flaxman

1.1 A meta-analysis example

1.2 Another meta-analysis example

1.3 Summary

2. Statistical models for rates, ratios, and durations / Abraham D. Flaxman

2.1 A motivating example

2.2 Binomial model

2.3 Beta-binomial model

2.4 Poisson model

2.5 Negative-binomial mode

2.6 Transformed normal models

2.7 Lower-bound data model

2.8 Quantification of uncertainty

2.9 Comparison

2.10 Summary and future work

3. Age pattern models / Abraham D. Flaxman

3.1 Definition of spline models

3.2 Choosing knots

3.3 Penalized spline models

3.4 Augmenting the spline model

3.5 Summary and future work

4. Expert priors on age patterns / Abraham D. Flaxman

4.1 Priors on level

4.2 Priors on monotonicity

4.3 Priors are not just for splines

4.4 Hierarchical similarity priors on age patterns

4.5 Summary and future work

5. Statistical models for heterogeneous age groups / Abraham D. Flaxman

5.1 Overlapping age-group data

5.2 Midpoint model

5.3 Disaggregation model

5.4 Midpoint model with group width covariate

5.5 Age-standardizing and age-integrating models

5.6 Model comparison

5.7 Summary and future work

6. Covariate modeling / Abraham D. Flaxman

6.1 Cross-walk fixed effects to explain bias

6.2 Predictive fixed effects to improve out-of-sample estimation

6.3 Fixed effects to explain variance

6.4 Random effects for spatial variation

6.5 Covariates and consistency

6.6 Summary and future work

7. Prevalence estimates from other data types / Abraham D. Flaxman

7.1 A motivating example

7.2 System dynamics model of disease in a population

7.3 Endemic equilibrium

7.4 Forward simulation examples

7.5 Summary and future work

8. Numerical algorithms / Abraham D. Flaxman

8.1 Markov chain Monte Carlo

8.2 The Metropolis-Hastings step method

8.3 The Adaptive Metropolis step method

8.4 Convergence of the MCMC algorithm

8.5 Initial values for MCMC

8.6 A meta-analysis example

8.7 Empirical Bayesian priors to borrow strength between regions

8.8 Summary and future work

8.9 Challenges and limitations

Section Two | Applications

9. Knot selection in spline models / Yong Yi Lee, Theo Vos, Abraham D. Flaxman, Jed Blore, and Louisa Degenhardt

10. Unclear age pattern, requiring expert priors / Hannah M. Peterson, Yong Yi Lee, Theo Vos, and Abraham D. Flaxman

11. Empirical priors / David Chou, Hannah M. Peterson, Abraham D. Flaxman, Christopher J.L. Murray, and Mohsen Naghavi

12. Overlapping, heterogeneous age groups / Mohammad H. Forouzanfar, Abraham D. Flaxman, Hannah M. Peterson, Mohsen Naghavi, and Sumeet Chugh

13. Dealing with geographical variation / Abraham D. Flaxman, Khayriyyah Mohd Hanaah, Justina Groeger, Hannah M. Peterson, and Steven T. Wiersma

14. Cross-walking with fixed effects / Amanda Baxter, Jed Blore, Abraham D. Flaxman, Theo Vos, and Harvey Whiteford

15. Improving out-of-sample prediction / Ali Mokdad, Abraham D. Flaxman, Hannah M. Peterson, Christopher J.L. Murray, and Mohsen Naghavi

16. Risk factors / Stephen S. Lim, Hannah M. Peterson, and Abraham D. Flaxman

17. The compartmental model / Sarah K. Wulf, Abraham D. Flaxman, Mohsen Naghavi, and Giuseppe Remuzzi

18. Knot selection in compartmental spline models / Marita Cross, Damian Hoy, Theo Vos, Abraham D. Flaxman, and Lyn March

19. Expert priors in compartmental models / Alize Ferrari, Abraham D. Flaxman, Hannah M. Peterson, Theo Vos, and Harvey Whiteford

20. Cause-specific mortality rates / Theo Vos, Jed Blore, Abraham D. Flaxman, Hannah M. Peterson, and Juergen Rehm

Conclusion / Abraham D. Flaxman, Christopher J.L. Murray, and Theo Vos

Appendix: GBD Study 2010 spatial hierarchy

References

Contributors

About the editors

Index