(Author: Roland Rashleigh-Berry
Date: 24 Aug 2015)
Introduction
This is a brief summary of the Spectre clinical trials reporting system.
All its components as well as its extensive documentation can be downloaded
from the main web site whose URL is below. You should refer to that web
site to gain a fuller understanding of the system.
Spectre has been made available to the public for educational purposes.
The reporting system is in everyday use by a pharmaceutical company. Although
it is in use, you should not assume that the system has been validated
for this purpose. If you use this system for clinical reporting purposes,
rather than just educational purposes, then you must comply with industry
regulations and your own Standard Operating Procedures for QC purposes.
Reporting "Engine"
Some programmers make the mistake of thinking about Spectre as a reporting
system. But I never wrote it intending it to be a reporting system. It
is a reporting engine. It is there to make simple the usual tasks
of calculating counts and percentages and calculating descriptive statistics.
Plus it has some basic calculations of p-values as an option. It creates
output datasets of this in the format you want (either transposed or not
transposed). You are supposed to report these datasets directly using the
Spectre macros or indirectly using your own "proc report" code or further
process the output dataset such as adding your own p-values in as observations
or new columns.
Above all, it was designed for speed. High speed. Designed to
cut through huge volumes of data in seconds rather than minutes.
Its job is to take in data and use CPU cycles to produce these datasets
at the maximum possible speed, rather like an automobile engine is designed
to take in gas and air and produce a rotary motion at high efficiency.
And just as this automobile engine can be fitted to multiple types of automobile
or even put to purposes other than to power an automobile, the Spectre
macros similarly don't have to be used in a particular way. So there is
nothing to "understand" about them and no philosophy about them that you
have to tune into to use them correctly. You use them as you see fit to
use them.
My aim in creating them with the high speed they work at was to create
a "push-button reporting system". I achieved that in about 2005 and reported
two clinical trials using that system, basically using a spreadsheet. Something
that in 2015, people are still striving for ten years after I achieved
it. I hope you will also reach that goal and if you are striving to reach
that goal then maybe I would be interested to help you. But if you want
to use these macros as part of a complex system or complex way of working
then go ahead and do it, but I would not want to become involved in those
efforts.
Features
Creates scripts to run dataset build and report production programs in
the correct order automatically
All runs are done in batch mode and output is pure text output (or PostScript
for figures)
Runs every SAS program independently to avoid one program affecting another
Has a system for handling titles and footnotes such that these can be kept
separate from reporting code modules
The titles and footnotes system has the potential to call reporting macros
Creates a log of every run with both start and stop times and the user
who launched the run
Creates a log of all outputs produced
Contains two powerful tools for creating safety tables which require minimal
parameters set
Output can be produced in more than one client style
Page labels of the general form "Page x of Y" are automatically added
Individual page margins can be specified for a study
Font size can be specified and the resulting line size and page size automatically
calculated
Extensive support is provided for producing bookmarked PDFs from the output
(Unix/Linux platforms only)
Contains a large number of shell scripts that help add power to the Unix
environment
Contains a collection of SAS macros that can be used to simplify writing
reporting modules
Has its own set of Unix learning materials to bring programmers up to speed
with the Unix environment
Study Reporting
The study reporting process flow is summarized on the page whose URL is
below. Reading it will give you an idea of what it is like to use Spectre
for study reporting.