Having emerged from pre-clinical testing (in the lab and animal models), a patented molecule (1) is ready for experiments in people. Two questions confronting a drug company never change: Is this chemical safe? And will it work? Clinical trials are designed to answer those questions in a rigorous, standardized way.
Drug companies pay doctors (2) , usually affiliated with university hospitals, to enlist patients (3) for three overlapping stages of research.
Phase I trials test the safety and pharmacokinetics of a compound on small groups of a dozen or so patients.
Phase II trials, which are slightly larger, establish optimal dosages and identify side effects.
Phase III trials investigate the efficacy and safety of a drug, often involving hundreds or thousands of individuals cared for by dozens of doctors in a variety of locations.
Combining two streams of information remains the major IT challenge. Some data are from laboratory machines (4) , which analyze bodily fluids and tissues. Using paper or computers, nurses and clerks also supply data about patients' experiences (5) . Both types of information are usually compiled manually and added to the study data set (6) . To guard against accidental (or deliberate) manipulation, the results are inspected by a data monitor (7) , who works for the drug company and ensures the study's rules are being followed.
When the doctors (2) are finished, the data set (8) is "locked," and sent to consultants for a statistical analysis. At this point, drug company executives (9) — who may be managers, physicians, chemists, biologists or marketing experts — face difficult decisions.
Any questionable data (10) might prompt more questions for the data monitor (7) . If a drug is harmful, or does not work as expected, the data are essentially discarded, i.e. filed in the company's archives (11) . This is by far the most frequent result of any clinical trial. If a drug seems promising but not as effective as the company hoped, it may warrant s an additional clinical study. With approval from the company, physicians running the study may publish their results in a peer-reviewed journal (13) . Only if the company hits the jackpot — if its drug is safe, effective, and likely to be profitable — will information be released in paper or electronic form to the FDA (14) , where a detailed review by government and academic experts may take months or even years.
Back to Clinical Trial Data Management: Tortured by Paper