A variety of biostatistics software suits are available around to complicate our choice of the software to downlaod, install and use. Wikipedia provides a very useful comparison page to help you make your choice. Of course, some packages like the free R programming language, and the commercial SASl, are still the best "professional" solutions, but their rather flat learning curve limits their usage by doctors and some professionals of biomedical research.

If you do not want to pay and you search a general purpose solution, then EpiInfo, AM Statistical Software, PSPP and MicroIris are some good choices. All of these programs need to be downloaded and installed before you are able to use them, and some of them are not simply portable. However, a new concept in this regard has emerged which is the on-demand software for statistics. Is fas as I know, the concept began with OpenEpi, a very useful and online statistical software provided freely to everyone. The software helps you do some calculations and especially learns you a lot of things about epidemiology and biostatistics without even needing to login. The source code is also available free of charge for your local use. I downloaded and installed the source code and it worked with no problem: already a good score for a downloadable online program.

Another, more complete solution is BrightStat (pay attention not to get hunted by BrightStats). The software is created by Dr Daniel Stricker from the University of Bern, Switzerland. It provides a clean and modern user interface based on Flash. The application itself is apparently written in PHP and MySQL. You may upload your data to the application and make some data management work on them. The software let's you do a variety of statistical tests, including descriptive analysis, Chi square, Binomial, ANOVA, Logistic regression, Kaplan Meier, etc. It also produces principal graphics, including Boxplots. I uploaded a database of 1600 records with 48 attributes each. The database was in a flat Excel sheet and was imported without any problem. The interface for data management and variable creation worked nicely, and descriptive analyses produced some nice looking reports. I tried some graphics works and it woked very well - even better than some installed counterparts. I congratulate the author of this software frankly for his clean work. I am confident the software will find good appreciation in medical community, provided that it is hosted on a more powerful server. In fact, for some more complex calculations it simply created time out errors. I suppose BrightStat is written in wholly in PHP. I am not sure if it is the best language to use for such a this software (I would suggest Python programming language or JSP). In anyway, this is a great work and I propose you to test it.

OpenEpi website: http://www.openepi.com/menu/openEpiMenu.htm

BrightStat website: http://brightstat.com/index.php

Until recently, the Oracle Corp's Oracle Clinical, and the Phase Forward's ClinTrial were the only big bosses of the clinical trial management system, with annual costs of tens of thousands of dollars. As a result only big pharmaceutical and biotechnology laboratories could afford such systems, which seem necessary for complying with international requirements of clinical trials.

OpenClinica, by Akaza Research company is in one of rare free and open source softwares available for the management of clinical trials. Another alternative, TrialDB is maintained and developed by a team at Yale University Medical Informatics department. Although copyrighted, it is freely available under GNU licence (which means that it cannot be used or modified for commercial use).

Both systems are web based: OpenClinical uses JSP and TrialDB uses ASP.NET (with Visual Basic.NET behind). They provide a database interface which is able to connect to an Oracle system, although theoretically it would be possible to connect them to any other database such as MySQL and PostgreSQL. TrialDB uses a Microsoft Access data base access internally for creating case report forms. It demands clients to have Internet Explorer as their web browser.

They implement workflows and provide different customizable access groups. Trial DB uses an entity-attribute-value (EVA) architecture which is somehow different from the conventional relational databases.

Although both systems provide modules for design and creation of case report forms (CRFs), the user interface of OpenClinica seems to be more modern and flexible than that of TrialDB for this purpose. Both systems have modules for defining access groups such as administrator, data coordinator, investigator, monitor, and data manager. The creation of annotations and queries is possible with both systems.

TrialDB supports access to a number of controlled vocabularies (e.g., ICD-10, DSM-IV, the Cerner/Multum Drug Lexicon, the NCI Common Toxicity Criteria) during data entry. However, it is more difficult to install and set up -as I understand from its website.

Both systems provide various export formats such as tabulated text, SPSS or SAS. OpenClinica provides also export in Clinical Data Interchange Standards Consortium's (CDISC) format.

In any way, these two sofwares are good initiatives for a new domain of open source projects. At the moment, I think they can also enjoy from some commercial success (especially for OpenClinica).

In a scientific and public health point of view, one should be aware that international requirements such as those demanded by the internationl conference for harmonisation (ICH) are so sophisticated that it has become almost impossible for the developing countries as well as for small and midium sized companies to carry out clinical trials without having a patent software. The arrival of these patented open source softwares is certainly welcomed and appretiated.

Open Clinical website: http://www.openclinica.org/

TrialDB website: http://ycmi.med.yale.edu/TrialDB/index.shtm

R prgramming language is a free and open source language and working environment for statistical computing and data mining, which is distributed under GNU licence. It has become the preferred language of academics in the domain of biostatistics and bioinformatics. The growing number of contributed packages have provided R with numerous functionalities. Some serious project such as bioconductor project, have further increased its penetration in the domain of genomics and biotechnology. More over, R graph gallery is proabably the most comprehensive collection of data visualisation funtionalities which have ever existed in a free or commercial statistical software.

However, companies are still working with SAS System of SAS Institute, which is a software framework for data entry, retrieval, management ad mining. A friend of mine has recently obtained his PhD in bioinformatics, and for finding a job in a company, he finally had to obtain a SAS certificate, although he had done all of his thesis working with R. Do companies have to choose SAS instead of R for finding clients or making projects?

For companies the situation is a little different. Let's see why a company may prefer to choose R as its statistical analysis software?

• It is available for no cost.
• Functions' source codes are open.
• There are more
• The software evolution is more rapid.
• Communitiy and especially academic support.

And why a company may prefer to use SAS?

• Its database is enhanced for large volumes of data.
• Data management is easier.
• The software is more user friendly.

If you are a company which manupulate data bases with mild to moderate turn over, R is a better choice for you. If you are a company with a large amount of data entry and data management, and thus with a large turn data over, you will need SAS. Enough documentation and support is now available in both cases. If you are already using SAS and you want to change for R to reduce your costs, do not forget to consider the cost of this migration, but in anyway such a change will reduce your costs.

If you want the software more for business analysis and less for biomedical computing, know that SAS is the leading company in the domain. However, a good R programmer can do also high quality business analysis with R.