David Schaffer’s Commentary
Brief outline of my activities developing the BLIPS software system
Phase I – 1974-1977 The George Washington University, Washington DC
Thomas A. Ban, editor: ECDEU Manual on Documentation and Assessment Procedures for Clinical Trials in Neuropsychopharmacology. Second Edition (1983)
I served mainly as a programmer implementing the already-designed system, Biometric Laboratory Information System II (BLIPS II), where the "II" was meant to distinguish it from an earlier system (I). The main difference between I and II was that the set of psychiatric instruments (assessment forms) supported by BLIPS I was small (~6). BLIPS II expanded the basic set of instruments and also supported almost any new instrument that an investigator in the Early Clinical Drug Evaluation Unit (ECDEU) (Guy, 1976) network wanted to add.
The BLIPS II system supported a standard documentation for the data collected with each instrument in a clinical trial. The first step of the system's operation was the reading of the data sheets ("mark sense" optical scan paper forms), and the running of a battery of error detection algorithms, to insure data quality. Error reports were communicated back to each investigator as rapidly as possible, knowing that opportunities for retrieving missing/wrong data might be transient. Once a data set was declared ready-to-analyze, a standard set of algorithms was run including a complete listing of the raw data, summary statistics (means and variances), cross tabulations and finally Analysis of Variance (ANOVA) and a narrative summary.
My activities focused mainly on implementing the analysis scripts for all the supported instruments and troubleshooting any processing difficulties that arose in the day to day activities of the Lab.
The staff was about a dozen people including top management (e.g., Roland R. Bonato and William Guy), several statisticians (e.g., Kenneth Yang, Patricia A. Cleary), several data clerks responsible for processing the studies as rapidly as possible (investigators were always keen to have their results back ASAP so they could publish their findings), the software staff (Luis Aguilar and me), and the operators (Louis Napper) for the IBM system 360 computer that was devoted exclusively to BLIPS because of the sensitive nature of the data we processed. We used punch-card technology in those days. Remote cathode ray tube (CRT) terminals were not yet common and the personal computer lay in the future.
There were two main objectives for BLIPS: 1) to provide a rapid standard analysis of each clinical trial in the ECDEU network of investigators so that the investigators, FDA and NIMH could be assured of a standard view of emerging research into psychopharmacological compounds; and 2) to create a growing database that might support future research into larger questions than just the effect of one compound in one cohort. Such questions might involve investigating classes of compounds in a given disease cohort, or the investigation of the broad effects of a compound or class of compounds in multiple cohorts.
The decision was taken in 1977 by NIMH, the sole funding source for the Biometrics Lab, not to continue the project. At that point, I chose to follow Bill Guy to the Tennessee Neuropsychiatric Institute (TNI) in Nashville, TN, and continue my work with BLIPS.
Phase II – 1977-1980 Tennessee Neuropsychiatric Institute, Vanderbilt University, Nashville, TN
Here I assumed responsibility for what we called the Biometric Lab of TNI. Our group was responsible for the timely collection of all the assessments (data) from all the clinical research ongoing at TNI, (excluding a Pharmacology unit under Fridolin Sulser). I was also the "keeper of the blinds," so was on call 24-7 in case an adverse reaction to a study compound required breaking the blind.
I had brought the BLIPS II system with me and got it up and running on the University’s main computer, a Digital Equipment DEC-VAX machine. My first challenge was that the preprocessing software would not port successfully to the VAX and, besides, we didn’t use the "mark sense" forms. So, I rewrote the entire “preproc” adding a few new data quality checks in the process. During this time, Jerry Levine (our program manager at NIMH) was on sabbatical in Pisa, Italy, and struggling to get the BLIPS II system operating on their computers. The main stumbling block was also the “preproc.”
So, in Dec 1977, Levine supported my visit there. With the very capable assistance of a young programmer, Gabriele Massimetti, we were able to install my software and get the system running. Although Gabriele spoke no English and I spoke no Italian, with a combination of IBM Job Control Language, Fortran and arm-waving, we managed to work together. We were supported at the Clinica Psychiatria at the University of Pisa by Giovanni B. Cassano and Luciano Conti.
I continued development of TNI_BLIPS, adding more forms and some additional analyses to the standard package. These enhancements were responsible for a second visit to PISA in 1981 to share the software.
I had had high hopes for the TNI chapter in my life to provide opportunities for my engineering skills to be used for things like modeling mental functioning and malfunctioning. The teams there were diverse and the promise was great. However, after three years little research of the multidisciplinary sort developed; I realized that my plan was somewhat foolish. It was unrealistic to expect a productive academic research career without a PhD, so in the Fall of 1980 I left TNI to pursue my PhD in the Electrical and Biomedical Department at Vanderbilt.
September 21, 2017