Advances in Statistical Methods for the Health Sciences

by Auget, Jean-louis; Balakrishnan, N.; Mesbah, Mounir; Molenberghs, Geert

• ISBN13: 9780817643683
• ISBN10: 0817643680
• Edition: 1st
• Format: Hardcover
• Copyright: 2006-12-30
• Publisher: Birkhauser

Summary

Statistical methods have become increasingly important and now form integral part of research in the health sciences. Many sophisticated methodologies have been developed for specific applications and problems. This self-contained volume, an outgrowth of an "International Conference on Statistical Methods in Health Sciences," covers a wide range of topics pertaining to new statistical methods and novel applications in the health sciences. The chapters, written by leading experts in their respective fields, are thematically divided into the following areas:* Prognostic studies and general epidemiology* Pharmacovigilance* Quality of life* Survival analysis* Clustering* Safety and efficacy assessment* Clinical design* Models for the environment* Genomic analysis* Animal health This comprehensive volume will be highly useful an of great interest to the health science community as well as practitioners, researchers, and graduate students in applied probability, statistics, and biostatistics.

Table of Contents

Preface	xix
Contributors	xxi
List of Tables	xxix
List of Figures	xxxv
PART I: PROGNOSTIC STUDIES AND GENERAL EPIDEMIOLOGY
Systematic Review of Multiple Studies of Prognosis: The Feasibility of Obtaining Individual Patient Data	3	(16)
D. G. Altman
M. Trivella
F. Pezzella
A. L. Harris
U. Pastorino
Introduction	3	(2)
Systematic Review Based on Individual Patient Data	5	(1)
A Case Study: Microvessel Density in Non-Small Cell Lung Cancer	6	(6)
Identifying studies (data sets) and obtaining the data	7	(2)
Checking the data	9	(2)
MVD measurements	11	(1)
Meta-analysis	12	(1)
Discussion	12	(7)
Systematic review of prognostic studies using individual patient data	13	(1)
The need for higher-quality prognostic studies	14	(2)
References	16	(3)
On Statistical Approaches for the Multivariable Analysis of Prognostic Marker Studies	19	(20)
N. Hollander
W. Sauerbrei
Introduction	19	(1)
Examples: Two Prognostic Studies in Breast Cancer	20	(1)
Statistical Methods	21	(2)
Regression models	21	(1)
Classification and regression trees (CART)	22	(1)
Formation of risk groups	23	(1)
Results in the GBSG-2 Study	23	(7)
Regression models -- standard applications	23	(2)
Regression models -- the MFP-approach	25	(1)
Summary assessment -- implication of the modelling strategy	25	(3)
Application of classification and regression trees	28	(2)
Formation and Validation of Risk Groups	30	(2)
Discussion	32	(7)
References	35	(4)
Where Next for Evidence Synthesis of Prognostic Marker Studies? Improving the Quality and Reporting of Primary Studies to Facilitate Clinically Relevant Evidence-Based Results	39	(22)
R. D. Riley
K. R. Abrams
P. C. Lambert
A. J. Sutton
D. G. Altman
Introduction and Aims	40	(2)
Prognostic Markers and prognostic marker studies	40	(1)
The need for formal evidence syntheses of prognostic marker studies	40	(1)
Aims of this chapter	41	(1)
Difficulties of an Evidence Synthesis of Prognostic Marker Studies	42	(7)
Poor and heterogeneous reporting	42	(2)
Poor study design and problems clarifying study purpose	44	(1)
Little indication of how to implement markers in clinical practice	45	(1)
Small sample sizes	46	(1)
Publication bias, selective within-study reporting, and selective analyses	46	(2)
Lack of appreciation or validation of previous findings	48	(1)
What Improvements Are Needed in Primary Prognostic Marker Studies?	49	(2)
Evidence Synthesis Using Individual Patient Data Rather than Summary Statistics	51	(3)
Discussion	54	(7)
References	55	(6)
PART II: PHARMACOVIGILANCE
Sentinel Event Methods for Monitoring Unanticipated Adverse Events	61	(14)
P. A. Lachenbruch
J. Wittes
Introduction	61	(2)
Examples	63	(1)
Usual Approaches to Monitoring Safety	64	(2)
Methods for Sentinel Events	66	(5)
Constant follow-up time	66	(2)
Censoring at a fixed calendar time	68	(3)
Methods for Sentinel Event Rates	71	(1)
Bayesian Models	72	(1)
Summary	73	(2)
References	74	(1)
Spontaneous Reporting System Modelling for the Evaluation of Automatic Signal Generation Methods in Pharmacovigilance	75	(20)
E. Roux
F. Thiessard
A. Fourrier
B. Begaud
P. Tubert-Bitter
Introduction	76	(2)
Methods	78	(6)
Spontaneous reporting system modelling	78	(1)
Exposure to the drug (Ti)	79	(1)
Events' relative risk (RRij)	79	(1)
Reporting probability (pij)	80	(4)
Data generation process	84	(1)
Application	84	(2)
Values of the model parameters	84	(2)
Application of the empirical Bayes method	86	(1)
Results	86	(2)
Discussion	88	(1)
Conclusion	89	(6)
References	90	(5)
PART III: QUALITY OF LIFE
Latent Covariates in Generalized Linear Models: A Rasch Model Approach	95	(14)
K. B. Christensen
Introduction	95	(1)
Generalized Linear Mixed Models	96	(2)
Latent regression models	97	(1)
Interpretation of Parameters	98	(1)
Generalized Linear Models with a Latent Covariate	98	(3)
Model	99	(1)
Interpretation of parameters	99	(1)
Parameter estimation	100	(1)
Example	101	(3)
Latent covariate	102	(1)
Job group level effect of the latent covariate	103	(1)
Discussion	104	(5)
References	105	(4)
Sequential Analysis of Quality of Life Measurements with the Mixed Partial Credit Model	109	(18)
V. Sebille
T. Challa
M. Mesbah
Introduction	110	(1)
Methods	111	(5)
The partial credit model	111	(1)
Estimation of the parameters	112	(1)
Sequential analysis	112	(1)
The Z and V statistics	112	(1)
Traditional sequential analysis	113	(1)
Sequential analysis based on partial credit measurements	113	(2)
The sequential probability ratio test and the triangular test	115	(1)
Simulation design	115	(1)
Results	116	(5)
Discussion	121	(1)
Conclusion	122	(5)
Appendix	122	(1)
References	123	(4)
A Parametric Degradation Model Used in Reliability, Survival Analysis, and Quality of Life	127	(12)
M. Nikulin
L. Gerville-Reache
S. Orazio
Introduction	127	(2)
Degradation Process	129	(1)
Estimation Problem	130	(1)
Linear MVUE for a	131	(1)
Solution of the Optimizatio Problem	132	(1)
Estimation of σ2 and θ0	132	(7)
References	134	(5)
Agreement Between Two Ratings with Different Ordinal Scales	139	(12)
S. Natarajan
M. B. McHenry
S. Lipsitz
N. Klar
S. Lipshultz
Introduction	139	(3)
Notation and Model	142	(2)
Examples and Interpretation	144	(3)
Discussion	147	(4)
References	147	(4)
PART IV: SURVIVAL ANALYSIS
The Role of Correlated Frailty Models in Studies of Human Health, Ageing, and Longevity	151	(16)
A. Wienke
P. Lichtenstein
K. Czene
A. I. Yashin
Introduction	151	(2)
Shared Frailty Model	153	(2)
Correlated Frailty Model	155	(1)
Correlated Gamma Frailty Model	156	(7)
Swedish breast cancer twin data	157	(1)
Parametric and semiparametric models	158	(1)
Correlated gamma frailty model with covariates	159	(1)
Cure-mixture models	160	(3)
Discussion	163	(4)
References	164	(3)
Prognostic Factors and Prediction of Residual Survival for Hospitalized Elderly Patients	167	(12)
M. L. Calle
P. Roura
A. Arnau
A. Yanez
A. Leiva
Introduction	167	(1)
Cohort Description and Follow-Up	168	(3)
Multistate Survival Model	171	(5)
Predictive process	173	(3)
Discussion	176	(3)
References	177	(2)
New Models and Methods for Survival Analysis of Experimental Data	179	(16)
G. V. Semenchenko
A. I. Yashin
T. E. Johnson
J. W. Cypser
Introduction	179	(1)
Semiparametric Model of Mortality in Heterogeneous Populations Influenced by Exogenous Interventions	180	(7)
The heterogeneous mortality model	180	(1)
Changes in the baseline hazard and frailty distribution	181	(2)
Semiparametric representation of the model	183	(1)
Interpretation of parameters	184	(3)
One Example of the Model Application to the Analysis of Poststress Survival Data	187	(3)
Heat shock of different duration applied to populations of nematodes Caenorhabditis elegans	187	(1)
Technical details	188	(1)
Results of fitting the model to the data	188	(2)
Discussion	190	(5)
References	191	(4)
Uniform Consistency for Conditional Lifetime Distribution Estimators Under Random Right-Censorship	195	(16)
P. Deheuvels
G. Derzko
Introduction and Results	195	(8)
Notation and statement of the problem	195	(3)
Definition of the estimators and uniform consistency	198	(5)
Proofs	203	(8)
Construction of the estimators	203	(3)
A useful auxiliary result	206	(2)
Concluding comments	208	(1)
References	208	(3)
Sequential Estimation for the Semiparametric Additive Hazard Model	211	(14)
L. Bordes
C. Breuils
Introduction	211	(2)
Example and Numerical Study	213	(2)
Assumptions and Theoretical Results	215	(5)
Technical Tools	220	(5)
Complete convergence, Anscombe condition, and exponential inequality	220	(1)
Technical results	221	(1)
References	222	(3)
Variance Estimation of a Survival Function with Doubly Censored Failure Time Data	225	(14)
C. Zhu
J. Sun
Introduction	225	(2)
Preliminaries	227	(1)
Variance Estimation	228	(3)
Simple bootstrap method	228	(1)
Generalized Greenwood formula	229	(1)
Imputation methods I and II	230	(1)
Numerical Results	231	(2)
Concluding Remarks	233	(6)
References	234	(5)
PART V: CLUSTERING
Statistical Models and Artificial Neural Networks: Supervised Classification and Prediction Via Soft Trees	239	(24)
A. Ciampi
Y. Lechevallier
Introduction	240	(1)
The Prediction Problem: Statistical Models and ANNs	241	(7)
Generalized linear models as ANNs	243	(2)
Generalized additive models as ANNs	245	(1)
Classification and regression trees as ANNs	246	(2)
Combining Prediction Models: Hierarchy of Experts	248	(2)
The Soft Tree	250	(7)
General concepts in tree construction	251	(1)
Constructing a soft tree from data	252	(1)
An example of data analysis	253	(3)
An evaluation study	256	(1)
Extending the Soft Tree	257	(2)
Conclusions	259	(4)
References	260	(3)
Multilevel Clustering for Large Databases	263	(12)
Y. Lechevallier
A. Ciampi
Introduction	263	(2)
Data Reduction by Kohonen SOMs	265	(4)
Kohonen SOMs and PCA initialization	266	(1)
Binning of the original data matrix using a Kohonen map	266	(2)
Dissimilarity for microregimens	268	(1)
Clustering Multilevel Systems	269	(2)
A two-level statistical model	270	(1)
Estimating parameters by the EM algorithm	270	(1)
Extracting Dietary Patterns from the Nutritional Data	271	(4)
References	273	(2)
Neural Networks: An Application for Predicting Smear Negative Pulmonary Tuberculosis	275	(14)
A. M. Santos
B. B. Pereira
J. M. Seixas
F. C. Q. Mello
A. L. Kritski
Introduction	276	(1)
Materials and Methods	277	(6)
Data set	277	(1)
Neural network design	278	(1)
Data selection for network design	279	(4)
Relevance of Explanatory Variables	283	(1)
Results	283	(1)
Conclusions	284	(5)
References	285	(4)
Assessing Drug Resistance in HIV Infection Using Viral Load Using Segmented Regression	289	(16)
H. Liang
W.-Y. Tan
X. Xiong
Introduction	289	(3)
The Model	292	(5)
The likelihood function	294	(1)
The prior distribution	294	(1)
The posterior distributions	295	(2)
The Gibbs Sampling Procedure	297	(1)
Analysis of the ACTG 315 Data	298	(3)
Conclusion and Dicussion	301	(4)
References	303	(2)
Assessment of Treatment Effects on HIV Pathogenesis Under Treatment By State Space Models	305	(18)
W.-Y. Tan
P. Zhang
X. Xiong
P. Flynn
Introduction	305	(1)
A Stochastic Model for HIV Pathogenesis Under Treatment	306	(3)
Stochastic differential equations of state variables	307	(1)
The probability distribution of state variables	308	(1)
A State Space Model for HIV Pathogenesis Under Antiretroviral Drugs	309	(1)
Estimation of Unknown Parameters and State Variables	310	(1)
An Illustrative Example	311	(3)
Conclusions and Discussion	314	(9)
References	317	(6)
PART VI: SAFETY AND EFFICACY ASSESSMENT
Safety Assessment Versus Efficacy Assessment	323	(12)
M. A. Foulkes
Introduction	323	(1)
Design Issues	324	(2)
Outcomes	324	(1)
Power	325	(1)
Population	326	(1)
Comparison	326	(1)
Analytic Issues	326	(3)
Compliance	326	(1)
Missing data	327	(1)
Confounding	327	(1)
Bias	328	(1)
Misclassification	328	(1)
Multiplicity	328	(1)
Analytic Approaches	329	(1)
Inferences	330	(1)
Conclusions	330	(5)
References	332	(3)
Cancer Clinical Trials with Efficacy and Toxicity Endpoints: A Simulation Study to Compare Two Nonparametric Methods	335	(14)
A. Letierce
P. Tubert-Bitter
Introduction	336	(1)
Setting	337	(3)
Method for the Simulation Study	340	(2)
Results	342	(4)
Conclusion	346	(3)
References	347	(2)
Safety Assessment in Pilot Studies When Zero Events Are Observed	349	(12)
R. E. Carter
R. F. Woolson
Introduction	349	(1)
Clinical Setting	350	(1)
Notation	351	(1)
Binomial Setting	351	(1)
Geometric Setting	352	(2)
Bayesian Credible Interval	354	(2)
Clinical Setting: Revisited	356	(1)
Summary	357	(4)
References	357	(4)
PART VII: CLINICAL DESIGNS
An Assessment of Up-and-Down Designs and Associated Estimators in Phase I Trials	361	(26)
H. K. T. Ng
S. G. Mohanty
N. Balakrishnan
Introduction	361	(1)
Notation and Designs	362	(3)
The biased coin design (BCD)	363	(1)
The k-in-a-row rule (KROW)	364	(1)
The Narayana rule (NAR)	364	(1)
Continual reassessment method (CRM)	364	(1)
Estimation of Maximum Tolerated Dose	365	(2)
Simulation Setting	367	(1)
Comparison of Estimators	368	(1)
Comparison of Designs	369	(18)
References	385	(2)
Design of Multicentre Clinical Trials with Random Enrolment	387	(14)
V. V. Anisimov
V. V. Fedorov
Introduction	387	(2)
Recruitment Time Analysis	389	(4)
Analysis of Variance of the Estimated ECRT	393	(2)
Inflation of the Variance Due to Random Enrolment	395	(1)
Solution of the Optimization Problem	396	(2)
Conclusions	398	(3)
Appendix	398	(2)
References	400	(1)
Statistical Methods for Combining Clinical Trial Phases II and III	401	(18)
N. Stallard
S. Todd
Introduction	401	(1)
Background	402	(5)
The clinical evaluation programme for new drugs	402	(1)
Combining clinical trial phases II and III	403	(1)
Background to sequential clinical trials	404	(3)
Methods for Combining Phases II and III	407	(6)
Two-stage methods	408	(1)
A multistage group-sequential method	408	(1)
A multistage adaptive design method	409	(1)
Example---a multistage trial comparing three doses of a new drug for the treatment of Alzheimer's disease	410	(3)
Discussion and Future Directions	413	(6)
References	415	(4)
SCPRT: A Sequential Procedure That Gives Another Reason to Stop Clinical Trials Early	419	(18)
X. Xiong
M. Tan
J. Boyett
Introduction	420	(1)
The SCPRT Procedure	421	(3)
Controlling the boundary	423	(1)
Boundaries in terms of P-values	423	(1)
An Example	424	(2)
SCPRT with Unknown Variance	426	(2)
Clinical Trials with Survival Data	428	(4)
Conclusion	432	(5)
References	432	(5)
PART VIII: MODELS FOR THE ENVIRONMENT
Seasonality Assessment for Biosurveillance Systems	437	(14)
E. N. Naumova
I. B. MacNeill
Introduction	438	(2)
Conceptual framework for seasonality assessment	438	(2)
δ-Method in Application to a Seasonality Model	440	(6)
Single-variable case	440	(1)
Two-variables case	441	(1)
Application to a seasonality model	442	(1)
Potential model extension	443	(1)
Additional considerations	444	(2)
Application to Temperature and Infection Incidence Analysis	446	(3)
Conclusion	449	(2)
References	450	(1)
Comparison of Three Convolution Prior Spatial Models for Cancer Incidence	451	(16)
E.-A. Sauleau
M. Musio
A. Etienne
A. Buemi
Introduction	451	(2)
Materials and Methods	453	(3)
ICAR model	455	(1)
Distance model based on the exponential function	455	(1)
Two-dimensional P-splines model	456	(1)
Implementation of the models	456	(1)
Results	456	(3)
Discussion	459	(8)
References	463	(4)
Longitudinal Analysis of Short-Term Bronchiolitis Air Pollution Association Using Semiparametric Models	467	(24)
S. Willems
C. Segala
M. Maidenberg
M. Mesbah
Introduction	467	(2)
Data	469	(1)
Sanitary data	469	(1)
Environmental data	469	(1)
Methods	470	(5)
Generalized additive models	470	(1)
Air pollution time series studies strategy	471	(3)
Criticism about the use of standard statistical software to fit GAM to epidemiological time series data	474	(1)
Results	475	(9)
Series of number of hospital consultations: Results with S-Plus	475	(7)
Series of number of hospital consultations: Results with R	482	(1)
SAS results	483	(1)
Discussion	484	(7)
References	486	(5)
PART IX: GENOMIC ANALYSIS
Are There Correlated Genomic Substitutions?	491	(28)
M. Karnoub
P. K. Sen
F. Seillier-Moiseiwitsch
Introduction	491	(1)
The Probabilistic Model	492	(4)
Parameter Estimation	496	(1)
New Test Statistic	497	(3)
Power Studies	500	(1)
Numerical Studies	501	(3)
Data Analysis	504	(3)
Discussion	507	(12)
Appendix 31.1	509	(1)
Appendix 31.2	510	(3)
References	513	(6)
PART X: ANIMAL HEALTH
Swiss Federal Veterinary Office Risk Assessments: Advantages and Limitations of the Qualitative Method	519	(8)
R. Hauser
E. Breidenbach
K. D. C. Stark
Introduction	520	(1)
Health Risks from Consumption of Milk and Dairy Products: An Example of a Qualitative Risk Assessment	521	(4)
Risk profile	521	(1)
Hazard identification	521	(1)
Risk network	522	(1)
Risk estimation	523	(1)
Recommendations for random sample planning and risk managers	524	(1)
Advantages and Disadvantages of Qualitative Risk Assessment	525	(1)
Statistics' Part in Qualitative Risk Assessment	526	(1)
References	526	(1)
Qualitative Risk Analysis in Animal Health: A Methodological Example	527	(12)
B. Dufour
F. Moutou
Introduction	528	(1)
Global Presentation of the Method	528	(2)
Qualitative Appreciation of the Probability of Each Event	530	(1)
Qualitative Risk Appreciation	531	(1)
Qualitative Appreciation Examples	532	(2)
Discussion	534	(5)
References	536	(3)
Index	539