Overview
Specific Aims
Importance
Background
Prelim. Studies

Methods
Study Design
Protocol
Data Collection
Analysis

Data
Introduction
Basic Data Sets
Variables Defs.
Codebook

Results
Working Papers
Publications
References

Introduction

A large amount of data were collected in the Seasons study. In nearly all cases, data were duplicate entered (or machine coded), and then processed so as to construct basic data sets. The process used to assemble these data sets is described in flow diagrams accompaning the description of the basic data sets.

Recruitment of Subjects

A total of 5000 subjects were recruited in total. Subjects were recruited from the Fallon Health Clinic in Worcester, or as part of additional minority recruitment effort in the Worcester area. From the initial contacts, a total of 1254 subjects (25.8%) met the verbal
eligibility criteria (spoke English) and consented to making a baseline appointment. Gender was recorded for these subjects (644 women, 610 men), but other demographic factors were not available. From this set of subjects, a total of 641 (51.1%) entered the Seasons study. The majority of subjects not entering the study either refused entry, or were judged ineligible (see Recruitment Flow Chart). Of the 1254 subjects initially eligible, a larger proportion of men entered the study (54.3%) than women (48.1%).

Basic 'Raw' Data

The basic data sets contain data on the 641 subjects who entered the seasons study (identified by the variable cstatus), plus some limited additional data on other subjects. Data includes simple questionaire responses, blood measurement data, and other 'processed' data, such as activity measures, dietary measures, and other scaled variables (such as the BECK Depression Index scales, the Social Desirability Scale, and the Seasonal Affective Disorders scale). Although we refer to these data as 'raw' data, the data have undergone considerable data cleaning. In addition, various processing has been completed, including creation of various scores, creation of summary nutrient and energy intake, and creation of summary activity measures in terms of metabolic equivalent scores.

Several data sets are described. Data sets differ by the definition of a basic record, as summarized below.

Data can be combined using linking variables. All data sets with the exception of the weather data set contain the study subject ID. The study protocol collected data every three months on diet, activity, light exposure and lipid values. These separate batteries of data are identified by a QUARTER number (ranging from 1 to 5) and indicating the sequence of quarterly collection.The date of lipid collection in a quarter is reported as the variable Q_DATE. For each quarter prior to the Lipid measurement, three telephone interviews were scheduled to record 24-hour recall of dietary intake, physical activity, and light exposure. These data can be linked to a subject in a quarter via the ID, QUARTER, and CALLDATE.

For some subjects, additional data were collected using an activity monitor (the Actilume). These data can be linked to subject's data via the ID and DATE.

Weather data are given for approximately 3 years over the course of the study.

Mixed models fit to 24-Hour data produce predicted values for use in regression models and are contained in the data set Pred. Vars.

In general, the full data are available upon request to the Principal Investigator, Dr. Ira Ockene. Example data sets on a subset of over 200 subject are available directly for exploritory purposes.

Assembling the Data

The process of assembling the data, while conceptually simple, was complicated. We review the process of assembling the data using flow-diagrams when presenting the basic data sets, but do not provide the basic more primitive data sets that were used in the process. The more primitive data sets can be made available upon request to the Principal Investigator, Dr. Ira Ockene.