Antonio Ciampi

Describing Health States: Methodologic Issues in Obtaining Values for Health States

In health status index construction quantitative values for different states of health are frequently obtained by presenting written descriptions to raters whose values are elicited using one or more methods. In this study the authors examined the influence of several aspects of this measurement process upon the quantitative results obtained. They prepared a set of written descriptions of health states, each state being described in both a standard point-form and a narrative format. The narrative format was written in the first person singular, and listed all symptoms or problems associated with the state, whereas the point-form description included only the most severe symptom or problem. Values for these states were elicited from a group of 64 patients using two commonly employed methods, the standard gamble of Von Neumann and Morgenstern and category rating. The results indicate that the type of scenario presented to the rater and the sequence in which the methods of assessment were used had a major influence on the results. This work indicates that there is a need to examine systematically the process of obtaining quantitative values before reliance can be placed upon the results.

RECPAM: a computer program for recursive partition and amalgamation for censored survival data and other situations frequently occurring in biostatistics. I. Methods and program features

The methodology of recursive partition and amalgamation in biostatistics is presented and a FORTRAN program for its implementation, RECPAM, is described. RECPAM can be used to obtain classifications of patients according to several criteria commonly occurring in clinical biostatistics: an example is prognostic classification based on survival data. Classes are defined by simple statements, expressed in clinical terms, about predictor variables (e.g. prognostic factors). Special features of RECPAM are: the possibility of implementing a variety of classification criteria, the integration of recursive partition and amalgamation, and the availability of several strategies for constructing classification trees. A simple example to illustrate input and output features is given. The scope and flexibility of RECPAM will be illustrated in greater detail in a subsequent paper.

Recursive Partition: A Versatile Method for Exploratory-Data Analysis in Biostatistics

In clinical and epidemiological studies, interactions among covariates are of primary importance. In the absence of clearly stated a priori hypotheses, it is usual to build predictive models by means of stepwise logistic or Cox regression, a practice that tends to emphasize main effects and overlook possible interactions.

Stratification by stepwise regression, correspondence analysis and recursive partition: A comparison of three methods of analysis for survival data with covaria

Abstract The problem of defining prognostic groups on the basis of censored survival times and covariates is central in medical biostatistics. Several methods have been proposed, but little is known about their relative advantages. Here three methods are discussed: Stepwise Regression, Correspondence Analysis and Recursive Partition. The approach is empirical in that the focus is on the performance on real data sets. Our example is discussed at length. We find that Stepwise Regression has the advantage of flexibility and economy of description, but is limited in discovering interactions and other complex features of the data. Correspondence Analysis is equally flexible, though less economical, and is very powerful in revealing unexpected features of the data: it is recommended as an exploratory tool. Recursive Partition is efficient in discovering interactions within large data sets and has the advantage of being the only method that produces clear descriptions in direct clinical terms; its flexibility, however, is limited, especially when the number of covariates is large relative to the number of individuals. Since no method is universally preferable, their joint use is recommended. A variety of criteria for ranking stratifications are proposed when a choice to be made.

Regression and recursive partition strategies in the analysis of medical survival data

Regression and clustering methods have both been used to explore the effects of explanatory variables on survival times for patients with cancer or other chronic diseases. This paper discusses effective and computationally feasible approaches for this task in situations where there are fairly large and complex data sets; the techniques stressed are all-subsets regression and a kind of recursive partition clustering. We compare the two approaches in a rather general way, in part by examining some survival data for patients with ovarian carcinoma, and conclude that both have strong points to recommend them.

Strategies for dietary intervention studies in colon cancer.

As a result of many studies in descriptive and analytic epidemiology, in animal carcinogenesis, and in the direct examination of body fluids for mutagens/carcinogens, it is possible to develop a list of dietary factors that may be associated with the high rate of colon cancer and related cancers in Western countries. This paper is concerned with the design of intervention studies to clarify which of these factors is important.

An approach to classifying prognostic factors related to survival experience for non‐Hodgkin's lymphoma patients: Based on a series of 982 patients: 1967–1975

The survival experience of 982 non‐Hodgkins lymphoma patients registered at Princess Margaret Hospital, Toronto, between 1967 and 1975, was studied. Prognostic groups were obtained by means of a classification procedure based on standard statistical techniques; the variables utilized in the classification were ones of known reproducibility which could be measured with little inconvenience to the patient. The results show that these prognostic groups give better information than the Ann Arbor staging classification in the sense that the survival curves are clearly separated and “good prognosis” and “poor prognosis” groups are clearly identified. Implications for therapy planning are briefly discussed.

A family of proportional- and additive-hazards models for survival data.

A family of proportional- and additive-hazards models for the analysis of grouped survival data is developed. This family generalizes the unpublished work of F.J. Aranda-Ordaz and follows Holford (1976, Biometrics 32, 227-237). It contains the proportional-hazards model for grouped data, as well as additive-hazards models with time trends. The time trends prove to be useful in an example in which the hazards of the two groups cross.

The prognostic value of HLA phenotypes in Hodgkin's disease.

Analysis of a group of 79 patients with previously untreated Hodgkins disease, whose HLA phenotypes were determined in 1972–73, shows that patients in a sub‐group with the specificities collectively known as Aw19 have significantly poorer survival than patients without these specificities. The degree to which the serologically‐detectable HLA antigens assist in determining the prognosis for survival has been analyzed using multiple regression analysis and retrospective stratification. Both approaches indicate that the association of HLA phenotype with survival is not accounted for by correlations between HLA antigens and other know prognostic factors, such as stage, histology, age, and sex. The greatest prognostic value of Aw19 appears to be for patients with relatively unfavorable age (>40 years), stage (III and IV), or histology (lymphocyte depletion, mixed cellularity).

Multi-Type Galton-Watson Process As A Model For Proliferating Human Tumour Cell Populations Derived From Stem Cells: Estimation of Stem Cell Self-Renewal Probabilities In Human Ovarian Carcinomas

Abstract. A mathematical model for proliferation of tumour cell populations is developed. the cell population is assumed to be organized in a hierarchy of decreasing proliferative potential and increasing degree of differentiation. Using some elements of the theory of Multi‐type Galton‐Watson processes, a method is proposed for the estimation of Psr, the probability of self‐renewal of tumour stem cells, from the experimental distribution of clonal unit sizes obtained in cell culture studies. Six data sets from patients with advanced adenocarcinorna of the ovary are used to demonstrate the method. Reasonable estimates are obtained, and the theoretical colony size distributions predicted by the model appear to be in good qualitative agreement with the experimental ones, and lend support to a stem cell model of tumour growth. the possible significance of Psr as a prognostic factor is briefly discussed.