Cost savings from database studies

Surveys and clinical trials can prove expensive and time consuming. In some cases, cross-sectional and longitudinal data analyses may provide the answers.

Studies conducted using commercially available population-based data sources offer several strategic advantages to companies. Database studies can be used for:

Early development: Identifying areas of unmet need; prioritising target product profiles; market size projections

Clinical trials planning: Affects of inclusion/exclusion criteria; expected recruitment rates

Safety: Background information on the disease; rates of comorbidities associated with the disease and various therapies; assistance with risk management planning

Commercial messages: Inputs for economic models; addressing key commercial messages (such as the prevalence of condition X)

This has led to the development of longitudinal data resources and both internal and external niche groups within companies focused on the analyses of these sources to address key company needs.

Cross-sectional and longitudinal data analyses can often represent a more cost-effective solution to address key needs within the company compared with conducting more expensive surveys and clinical trials. Often, companies can obtain the answer in a relatively short time frame rather than the months to years of a clinical trial or survey. In addition, safety signals can be evaluated as soon as a sufficient treated population is available.

There are many database resources, including:

• Government-sponsored population surveys (US National Health and Nutrition Examination Survey, UK Household Survey)
• Government statistical surveys (National Hospital Discharge Survey, UK Hospital Episode Statistics)
• Data associated with billing records or claims, either via claims providers (Canada’s Saskatchewan database, US PharMetrics claims warehouse) or employer based (MedStat) or government (US Medicaid and Medicare programmes, VA data)
• Data associated with electronic medical records (General Practice Research Database)

Larger companies may be able to build internal capabilities for use of all of these types of data sources, but there are many contract research organisations (CROs) and other types of research organisations available to assist. As a result, companies may acquire the data and conduct the study completely in-house or partner with a CRO to obtain the data and conduct the analyses for a specific study. For some data sources, all analyses must be contracted with the data source.

Options for partners are wide-ranging. Often the larger, longitudinal database vendors have research organisations within their team who can assist. Many CROs and academic institutions also have the necessary expertise.

Database analyses are not clinical trials and require knowledgeable personnel to properly conduct them. Key competencies involve a history of publications for both the data source and for those conducting the analyses; a staff comprised of epidemiologists, economists, statisticians and analysts; experience with the data source; and organisational knowledge of the regulatory, confidentiality and IT issues associated with the use of these sources, particularly the larger longitudinal sources, which can be hundreds of gigabytes in size.

Use of this data requires an understanding of why the data source exists. The data is produced for a variety of reasons. In the case of government-related surveys, there is a need to understand information on the health of the population as well as information about the health-care system in general which can be used for many purposes beyond just the primary objective. In the case of claims data, the need to have detailed billing records for accounting purposes facilitates the use of the same records for research purposes. In the case of the electronic medical record provided by GPs, the software was primarily designed to help practitioners manage their patients’ care.

Database studies cannot address all questions. For example, for some conditions not all patients meeting the symptom profile go to see their physician. In these cases, only population-based surveys using appropriate questions can answer the question. However, there are many situations where a study in one or a combination of data sources can rapidly provide information to inform key business needs.