|
|
|
|
|
|
|
This database will probably take several years to settle down, but it is one of the first attempts to base cost estimates on real data from across the industry. Some companies are already learning to apply their own correction factors to PICAS data. |
|
|
|
|
|
|
|
|
1.
Cracking the Patient Supply Problem |
|
|
|
|
|
|
|
|
Probably the most critical estimate needed is the rate of patient accrual, and here there is the opportunity for substantial progress. Some clinical study sites are beginning to organize their patient records far more effectively. This is a trend starting in the U.S. [7], with organizations offering databases of patients who can be matched to protocol selection criteria. The patients can be contacted directly and screening clinic visits scheduled very quickly. Such advertising may raise questions in Europe, but whatever the ethics, this shows how technology can change what has been traditionally the clinical research manager's nightmarefinding enough valid patients. |
|
|
|
|
|
|
|
|
2.
The Cautious Approach. |
|
|
|
|
|
|
|
|
The availability of such volumes of data enables more widespread application of a hugely underused method, the feasibility study or pilot study. The former is really a paper (now electronic) exercise, where we can simulate the clinical trial [8] before committing to major cost. Previously, such studies have been of limited value because the data source was unreliable or nonexistent, but now the world is changing. Interestingly, the concept may be more easily applied in general practice than in hospital medicine, because of the widespread use of computerized patient records in the former (at least in much of Europe). In the U.K., hospital patient records are mostly paper, and commonly fragmented and disparate. Only in dedicated clinical research units does one find a serious attempt at organizing patient records for ready access. There is, though, a basis for optimism, with U.K. hospital trusts now processing data on admissions, referrals, and outcomes, mainly for marketing reasons. Reasons for referral are coded using much the same dictionaries as are used by the drug industry in biometrics. So it is not too much of an extrapolation to see such codes helping to identify patients eligible for a particular study. However, this presupposes a fundamental change in the role of research among health care providers, of which more a little later. An interim approach may be to issue cheap pocket computers to clinicians, into which they can enter basic patient information during each consultation. Quality assurance of such data may be difficult but should provide much better guidance on referral rates than is generally available at present. |
|
|
|
|
|