American Journal of Epidemiology Vol. 155, No. 8 : 776-777
Copyright © 2002 by The Johns Hopkins University School of Hygiene and Public Health
BOOK REVIEWS |
Computer Programs for Epidemiologists: PEPI Version 4.0
Division of Epidemiology School of Public Health University of California at Berkeley Berkeley, CA 94720
Although conventional statistical software packages are adequate when you have a data set to work with, they are not always helpful when you need to do keyboard entry of data and rapidly perform simple analyses. For instance, you may want to replicate some analyses from a journal article and compute a Mantel-Haenszel odds ratio, or you may want to compute the sample size for your study while writing a grant proposal. Maybe you want to demonstrate to your students the impact of increasing sample size on the confidence intervals of a proportion. Perhaps you are a student and would like to do your epidemiology or biostatistics homework with some easy-to-use analytical routines. Wouldn't it be neat to have a package that can do all of these without a fuss? Epi Info version 6 (1
) had this great set of programs called EPITABLE that could do all sorts of analyses from computing kappa values to sample size. Sadly, Epi Info 2000 (2) has done away with it. STATA version 7 (3) has a suite of programs called Epitab that are designed for epidemiology and can handle keyboard data entry. However, anyone who has tried doing stratified analysis by keyboard entry in STATA version 7 will know that it is anything but easy! It is in this niche area that PEPI scores!
PEPI is an acronym for Programs for EPIdemiologists. This set of discrete programs, now in its fourth version, has 43 easy-to-use programs for epidemiologic analyses. The idea is to complement—not replace—conventional statistical packages. All but the logistic regression programs require keyboard entry. The programs cover an amazing array of applications: logistic regression, multiple significance tests, attributable fractions, meta-analysis, confidence intervals, direct and indirect standardization, smoothing of curves, life table and survival analyses, correcting for misclassification, and power and sample size estimations, to name just a few. To our knowledge, none of the conventional packages can perform such a variety of tasks without a data set. The software itself is free and can be downloaded from the publisher's Web site. Like Epi Info, it can be freely copied and distributed for personal use. The software can also be bought, along with a well-written manual. Abramson and Gahlinger have made a good effort to explain each of the programs with illustrative examples and exhaustive references. They have provided the formulas used for computations or at least provided the appropriate references. Unlike many manuals, they occasionally provide a brief discussion on controversies and debates surrounding the use of some of the analytical methods.
We put PEPI through a road test and found the programs to be simple and easy to use. The screen instructions were fairly clear and, with a Pentium processor, computational speeds were very fast. PEPI, however, has its limitations. First is the lack of graphical output. Second is the inability to store data in most modules; the user must create data files in ASCII, with variable names entered manually in a separate dictionary file (file imports from Epi Info and EpiData are possible). Our biggest disappointment, however, was that all the programs in PEPI, except one, are DOS based! Although the programs can run on Windows, once you open PEPI from the Windows START menu, it's your typical, outdated DOS interface. The user interface and the quality of the output leave much to be desired. If you prefer to use a mouse, sadly, it will be of no use to you much of the time. Using DOS-based programs in 2002 is painful indeed, and we wish the authors had converted all the programs into the Windows format. WHATIS, the only Windows program, is our pick for the best program in PEPI. If all the programs could be converted into the WHATIS type of format, PEPI will be a truly outstanding package! To their credit, the authors are making an effort to do this. Two more programs have been converted and updated, and these are now available for trial on this Web site: www.myatt.demon.co.uk/index.htm.
While the creators of PEPI are upgrading to the Windows platform, why not create a personal digital assistant (PDA) version? Given how nimble and versatile PEPI already is, a PDA version could significantly enhance every epidemiologist's and researcher's ability to "think on his/her feet" at meetings, in the classroom, or in the field. Another area where improvements can be made is by providing readymade 2 x 2 tables or blank cells for data entry. Statcalc in Epi Info is a great example of this. When a 2 x 2 table pops up on the screen with the exposure and outcome axes already labeled, chances of getting your rows and columns mixed up are remote. This is particularly important given the confusion that prevails in the way 2 x 2 table axes are labeled in epidemiology textbooks and statistical packages. It would also help if PEPI had an opening screen with a main menu pick list that shows program names and a brief description about each. The opening menu in Epi Info 2000 is a good model for this.
Will you need all the programs in PEPI? Probably not. We have, for example, never used the Jonckheere Terpstra test for trend or the Kullback-Leibler distances. However, more is good, as long as it does not overload your system and comes free! PEPI will take up around 3.6 megabytes on your hard disk, and that isn't too much these days. Some programs, we think, are too simplistic for comprehensive data analysis. For example, we would not do logistic regression with PEPI. Conventional packages are much more sophisticated for such analyses. Likewise, we would prefer specialized meta-analysis packages with graphical capabilities, like RevMan (4), to PEPI for performing meta-analyses. Given the increasing number of meta-analyses being published, future versions of PEPI would benefit from an expanded treatment of such techniques.
In sum, PEPI is the epidemiologist's Swiss army knife—a great set of tools, particularly for learning and teaching epidemiology. We plan to use it for teaching in an epidemiology methods course. PEPI facilitates a ready understanding of important epidemiologic concepts, unfettered by the complexities of statistical programming. For example, the impact of misclassification in a case-control study can be immediately demonstrated in a classroom setting without having to retire to the computer lab. If you know what you're looking for, this program can take you there promptly and without unnecessary fanfare. PEPI cannot replace your conventional package, but it can be a valuable help to supplement your everyday statistical computing needs.
NOTES
By J. H. Abramson and Paul M. Gahlinger
ISBN 0-9703130-2-0, Sagebrush Press, Salt Lake City, Utah (Telephone: 801-350-0237, E-mail: info{at}sagebrushpress.com, Web site: http://www.sagebrushpress.com), 2001, 305 pp., $59.95
REFERENCES
- Dean AG, Dean JA, Coulombier D, et al. Epi Info version 6: a word-processing, database, and statistics program for public health on IBM-compatible microcomputers. Atlanta, GA: Centers for Disease Control and Prevention, 1995.
- Dean AG, Arner TG, Sangam S, et al. Epi Info 2000, a database and statistics program for public health professionals for use on Windows 95, 98, NT, and 2000 computers. Atlanta, GA: Centers for Disease Control and Prevention, 2000.
- Stata Corporation. Epitab—tables for epidemiologists. In: STATA 7 reference manual. Vol 1. A–G. College Station, TX: Stata Press, 2001:442–91.
- Review Manager (RevMan) version 4.1 for Windows. Oxford, England: The Cochrane Collaboration, 2000.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  W. Winkelstein Jr. Vignettes of the History of Epidemiology: Three Firsts by Janet Elizabeth Lane-Claypon Am. J. Epidemiol., July 15, 2004; 160(2): 97 - 101. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||