Clinical databases

Yazdanpanah et al. (2002) calculated the resource use and cost for different stages of HIV infection in France based on a clinical database of HIV-infected patients between 1994 and 1998. The total costs attributable to bed-day and day-care inpatient care included the mean cost of each inpatient day times the length of stay, as well as total number of laboratory tests, dosage and quantity of medications, and total number of procedures. The total cost attributable to outpatient care included the mean physician and nurse fees per visit, as well as total number of laboratory tests and total number of procedures. In the absence of an AIDS-deflning event, the average total cost of care ranged from US 797 per person-month in the highest CD4 stratum to US 1,261 per person-month in the lowest CD4 stratum. 

Finally, there is the very important annotated CRF, which shows you where the variables in the clinical database come from on the CRF. The following is an example of an annotated medical history CRF page ... 

In order to reduce unnecessary data queries, the statistics group should be consulted early in the clinical database development process to identify variables critical for data analysis. Optimally, the statistical analysis plan would already be written by the time of database development so that the queries could be designed based on the critical variables indicated in the analysis plan. However, at the database development stage, usually only the clinical protocol exists to guide the statistics and clinical data management departments in developing the query or data management plan. 

There is only one good solution to handling free-text variables that are needed for statistical analysis. The free-text variables need to be coded by clinical data management in the clinical database. If the adverse events were coded with a dictionary such as MedDRA, the previous example might look like Program 2.3. 

The FDA may mandate specific postmarketing surveillance programs when a concern about the safety or efficacy of a new medication is suggested by either the preclinical or clinical database used for marketing approval. 

In a recent article, the FDA stated For excipients with a history of use, it may be possible to adequately address some or all of the safety issues through citation of the existing nonclinical and clinical database, marketing history, or regulatory status of the compound (e.g., GRAS status as a direct food additive may adequately support oral administration of that compound up to the levels permitted in foods) (25). Therefore, it appears reasonable that the FDA would consider the use safety data based on the food additive regulatory status of an excipient to evaluate the safety of the excipient. If a new excipient has undergone a food additive safety review, this may reduce the perceived risks associated with the development and use of a new excipient. 

For the appropriate development of covariate distribution models, the pharmaceutical industry has huge amount of data in their clinical databases. In addition, there are also public databases available which can be used, like the Congestive Heart Failure Database (http //www.physionet.org/) derived from patients undergoing cardiac catheterization at Duke Medical Centre during 1990-1996 (about 4000 patients, data on demographics, risk factors histories, cardiac catheterization, EKG, cardiac scores, follow-up data). 

Once the corrected copies are received, the data manager makes the change in the clinical database. An electronic audit trail is maintained in the clinical database of all data entered and changed. This audit trail tracks the date and time stamp and the identification of the person making the entry correction or change. 

In addition to the discrepancies generated as a result of study definition (univariate discrepancies), discrepancies may also arise when a batch validation detects data inconsistencies (univariate and multivariate discrepancies). Discrepancies are also identified by a visual review of the data, e.g., monitoring lists, SDV review. Discrepancies may also be created by people responsible for data analysis (e.g., statisticians, pharmacoeconomists, clinical pharmacologists). All discrepancies and data fields requiring verification or clarification are tracked using the clinical database. 

For example, metrics involving patient enrollment, visits, forms flow, and discrepancies may be tracked using the clinical database. An example of a tracking report for time from patient visit to receipt in-house is given in Table 2. 

A well-designed data management system typically will focus on the primary objective to facilitate the collection and cleaning of clinical data. Although it must also support analysis and reporting, it is not always possible to achieve an equal balance across all these requirements therefore, data are usually analyzed outside of the clinical database. 

Hospital databases may, however, be used for the establishment of reference values that are fuUy concordant with the IFCC recommendations." The requirement is that laboratory data be combined with information stored in clinical databases (i.e., to apply a direct sampling strategy instead of the distribution-based indirect method). Laboratory results are to be used as reference values only if stated clinical criteria are fulfilled. One may define criteria for selecting individuals who have a specified state of health or the disease for which reference data are necessary. Usually certain constraints are also imposed on the use of their laboratory results, such as allowing only one result of each analyte under study from each selected individual. Such reference values have one advantage over those based on direct sampling from other types of populations hospital-based reference values are ideal for the interpretation of results from hospitalized patients because they are produced under similar conditions. 

However, when anything new is discovered about a drug in phase IV, then, by definition, it will not be in the product label. Furthermore, sometimes, when such a signal is observed, a retrospective trawl through the preclinical and clinical databases can often uncover consistent information whose significance had not been earlier realized. In this case, a gap exists between what is known about a drug and what information has been provided to prescribers. 

In the previous GCP guidelines, the auditor s activity was confined to the company. The auditor used the CRF as a source document and checked the clinical database and final report against it. 

Time is the fundamental predictor in PK/PD models and therefore deserves some special attention. In NONMEM it is possible to specify the time points for observations and dosing events in the form of date and clock time. This is very convenient as this is the form in which the data is often stored in clinical databases. These dates and clock times are converted to decimal times in the preprocessing stages of the execution of a NONMEM run, and it is these decimal times that are used by NONMEM in the minimization procedure and that are provided in the tabulated output. From a plotting point of view, dates and clock times are not easy to work with. Except for cases with diurnal variations and/or annual rhythms (2,3), the extra dimension offered by dates and clock times are unnecessary and may actually make it harder to visualize the data in an informative way. 

The medical literature is perhaps the best indexed and most readily accessible of all information types in science and technology. The National Center for Biotechnology Information (NCBI) at the National Library of Medicine (NLM) was a leader in creating an online, high-quality clinical database to complement its venerable multi volume Index Medicus. NLM s... 

MEDLINE database, known as PubMed on the NLM platform, is available free worldwide, and is considered the premier source for searching the clinical medical literature back to 1966. Embase is the other key clinical database, produced by Elsevier and known for its international scope and exceptional coverage of the drug literature. 

Due to the limitations of the size of the clinical database, the evaluation of safety data and, therefore, risk characterization have long not been in the focus of statistical considerations. The evaluation of safety data is only briefly... 

---