Introduction The degree of generalisability of patient databases to the general population is important for interpreting database research. This report describes the representativeness of The Health Improvement Network (THIN), a UK primary care database, of the UK population. Methods Demographics, deprivation (Townsend), Quality and Outcomes Framework (QOF) condition prevalence and deaths from THIN were compared with national statistical and QOF 2006/ 2007 data. Results Demographics were similar although THIN contained fewer people aged under 25 years. Condition prevalence was comparable, e.g. 3.5% diabetes prevalence in THIN, 3.7% nationally. More

Objectives To build and test a model for the collection of computerised retrospective primary care data from the UK, and to assess its quality for use in medical and pharmaceutical research. Design Collection and evaluation of sampled retrospective general practice data recording. Setting General practices, using the Vision practice management software in the UK. Main outcome measures Quality indicators of completeness of data recording. Results Initial audit of 236 practices indicated good recording of prescribing in all practices and a high level of completeness of recording of clinical information in many of the practices. Conclusions In the group of practices studied, levels of recording were generally assessed to be of sufficient quality to enable a database of qualityevaluated, anonymised primary care records to be created.

No abstract

The use of healthcare databases in research provides advantages such as increased speed, lower costs and limitation of some biases. However, database research has its own challenges as studies must be performed within the limitations of resources, which often are the product of complex healthcare systems. The primary purpose of this document is to assist in the selection and use of data resources in pharmacoepidemiology, highlighting potential limitations and recommending tested procedures. This guidance is presented as a detailed text with a checklist for quick reference and covers six areas: selection of a database, use of multiple data resources, extraction and analysis of the study population, privacy and security, quality and validation procedures and documentation.

BackgroundLittle is known about the effects of human fetal exposure when a new drug is authorized unless it was specifically developed for use in pregnancy. Since many factors may contribute to adverse fetal effects, having comprehensive information about in utero exposures will enhance our ability to make correct determinations about causality.ObjectiveThe objective of the study was to assess the extent to which women, recruited without the intervention of health care professionals (HCPs), will provide information, suitable for research purposes, via the Internet or by phone on some potential risk factors in pregnancy.MethodsTo pilot direct-to-patient research for pharmacovigilance, we conducted a prospective, noninterventional study of medication use and lifestyle factors as part of the Pharmacoepidemiological Research on Outcomes of Therapeutics by a European ConsorTium (PROTECT) Consortium. Consenting women who self-identified as pregnant and residing in the United Kingdom (UK), Denmark (DK), The Netherlands, or Poland were recruited and could then choose to provide data every 2 or 4 weeks via the Internet or a telephonic interactive voice response system (IVRS). Self-reported drug use was compared with pharmacy register data in DK and with electronic health records in the UK.ResultsRecruited women were on average older and more highly educated than the general population. Most respondents chose a frequency of every 4 weeks (56.99%, 1177/2065). Only 29.83% (464/1555) of women with due dates occurring during the study provided information on pregnancy outcome. For those responding by Internet, over 90.00% (1915/2065) reported using >1 pregnancy-related medication, 83.34% (1721/2065) reported using >1 other medicine, and 23.53% (486/2065) reported only over-the-counter medications, not counting herbals and dietary supplements. Some respondents (7.16%, 148/2065) reported that they chose not to take a prescribed medication (mostly medicines for pain or inflammation, and for depression) and 1.30% (27/2065) reported using medicines that had been prescribed to a friend or family member (oxycodone, paracetamol, and medications for acid-related problems). Relatively few respondents reported using fish oil (4.60%, 95/2065), other dietary supplements (1.88%, 39/2065), herbal products (7.07%, 146/2065), or homeopathic products (1.16%, 24/2065). Most medications for chronic conditions that were listed in the Danish prescription registry were also self-reported (83.3%, 145/174 agreement), with larger discrepancies for medications indicated for short-term use (54.0%, 153/283 agreement) and pregnancy-related medications (66.1%, 78/118).ConclusionsSelf-reported information on medication use as well as other potential teratogenic factors can be collected via the Internet, although recruitment costs are not insubstantial and maintaining follow-up is challenging. Direct data collection from consumers adds detail, but clinical input may be needed to fully understand patients’ medical histories and capture birth outcomes.